{"owner":"the-open-agent","repo":"openagent","hasSkills":true,"totalSkillsCount":61,"totalTokensCount":63920,"categories":["claude-rule","anthropic-skill"],"hasMcp":false,"mcpConfig":null,"found":["CLAUDE.md","skills/1password/SKILL.md","skills/1password/references/cli-examples.md","skills/1password/references/get-started.md","skills/apple-notes/SKILL.md","skills/apple-reminders/SKILL.md","skills/bear-notes/SKILL.md","skills/blogwatcher/SKILL.md","skills/blucli/SKILL.md","skills/bluebubbles/SKILL.md","skills/camsnap/SKILL.md","skills/canvas/SKILL.md","skills/clawhub/SKILL.md","skills/coding-agent/SKILL.md","skills/discord/SKILL.md","skills/eightctl/SKILL.md","skills/gemini/SKILL.md","skills/gh-issues/SKILL.md","skills/gifgrep/SKILL.md","skills/github/SKILL.md","skills/gog/SKILL.md","skills/goplaces/SKILL.md","skills/healthcheck/SKILL.md","skills/himalaya/SKILL.md","skills/himalaya/references/configuration.md","skills/himalaya/references/message-composition.md","skills/imsg/SKILL.md","skills/mcporter/SKILL.md","skills/model-usage/SKILL.md","skills/model-usage/references/codexbar-cli.md","skills/nano-pdf/SKILL.md","skills/node-connect/SKILL.md","skills/notion/SKILL.md","skills/obsidian/SKILL.md","skills/openai-whisper-api/SKILL.md","skills/openai-whisper/SKILL.md","skills/openhue/SKILL.md","skills/oracle/SKILL.md","skills/ordercli/SKILL.md","skills/peekaboo/SKILL.md","skills/powerpoint/SKILL.md","skills/powerpoint/references/pptxgenjs.md","skills/sag/SKILL.md","skills/session-logs/SKILL.md","skills/sherpa-onnx-tts/SKILL.md","skills/skill-creator/SKILL.md","skills/slack/SKILL.md","skills/songsee/SKILL.md","skills/sonoscli/SKILL.md","skills/spotify-player/SKILL.md","skills/summarize/SKILL.md","skills/taskflow-inbox-triage/SKILL.md","skills/taskflow/SKILL.md","skills/things-mac/SKILL.md","skills/tmux/SKILL.md","skills/trello/SKILL.md","skills/video-frames/SKILL.md","skills/voice-call/SKILL.md","skills/wacli/SKILL.md","skills/weather/SKILL.md","skills/xurl/SKILL.md"],"skills":{"CLAUDE.md":"# CLAUDE.md\n\nThis file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.\n\n## Commands\n\n**Backend (Go)**\n```bash\ngo build # build backend binary\ngo run main.go # run backend (port 14000)\ngo test ./... # run all tests\ngo test ./object/... -run TestFoo # run a single test\n```\n\n**Frontend (React + shadcn/ui, built with Vite)**\n```bash\ncd web && yarn install\nyarn start # dev server on port 13001, proxies /api to localhost:14000\nyarn build # production build (vite writes build-temp, mv.js swaps it into web/build)\nyarn typecheck # tsc --noEmit\nyarn lint # eslint\n```\n\n`web-old/` holds the previous Ant Design frontend. It is kept for reference only\nwhile bugs are chased down against it — it is not built, served or linted, and it\ndoes not get new work.\n\n**Full production build** uses `build.sh` which cross-compiles for linux/amd64, linux/arm64, linux/riscv64.\n\nBackend serves the frontend as embedded static files (`embed.go`); during development the frontend proxy is configured to hit `localhost:14000`.\n\n## Architecture\n\n### Backend (Go / Beego)\n\n**Module:** `github.com/the-open-agent/openagent`\n\nThe backend is a standard [Beego](https://beego.vip/) MVC app. Every entity follows the same three-layer pattern:\n\n1. **`object/.go`** — struct definition (xorm tags for DB), and all DB access functions: `GetGlobal`, `GetCount`, `GetPagination`, `Get`, `Add`, `Update`, `Delete`. The primary key is always `(Owner string, Name string)`. Use `util.GetId(owner, name)` / `util.GetOwnerAndNameFromIdWithError(id)` for composite key serialization.\n\n2. **`controllers/.go`** — Beego controller methods wired to routes. Standard set: `GetGlobal`, `Get`, `Get`, `Add`, `Update`, `Delete`. Use `c.IsAdmin()` / `c.IsGlobalAdmin()` / `c.RequireSignedIn()` for auth. Paginated list APIs accept `p`, `pageSize`, `field`, `value`, `sortField`, `sortOrder` query params and use `pagination.SetPaginator`.\n\n3. **`routers/router.go`** — `beego.Router` calls grouped by entity (all routes for the same entity's CRUD kept together). No alphabetical ordering within a group. New entity routes should be inserted next to their entity's existing routes (not appended to the end of the file).\n\n**DB schema** is auto-migrated via `object/adapter.go` → `createTable()` using `engine.Sync2(new(EntityStruct))`. Add new entities there.\n\n**i18n (backend):** `i18n/locales/{en,zh}/data.json` — keys are namespaced by category (e.g. `\"comment:...\"`, `\"general:...\"`). Use `c.T(\"namespace:key\")` in controllers.\n\n### Frontend (React + shadcn/ui on Tailwind, TypeScript, Vite)\n\n**Routing** lives in `web/src/App.tsx`; every file in `web/src/pages` is its own lazy chunk, loaded by name. To add a new admin page:\n1. Create `web/src/pages/Page.tsx` (default export).\n2. Add a `` in `App.tsx` pointing at `page(\"Page\")`.\n3. Add a menu entry in `getNavGroups()` (`web/src/lib/nav.ts`), in the correct group — Basic / Connectors / Admin etc.\n4. Add the nav key to `web/src/components/common/NavItemTree.tsx` so it appears in the site Navbar Items config.\n\n**Page pattern:** list pages render `CrudListPage` (`web/src/components/crud/`) with a `ColumnDef[]`; edit pages render `SimpleEditPage` with an `EditField[]`, or compose `EditPageShell` directly when the layout is custom. Both patterns are illustrated by `MessageListPage.tsx` / `MessageEditPage.tsx`.\n\n**Backend API layer:** `web/src/backend/Backend.ts` — thin `fetch()` wrappers, one file per entity. Standard exports: `getGlobal`, `get`, `update`, `add`, `delete`.\n\n**Per-instance config:** the backend sends the Casdoor issuer, client ID and branding in the `jsonWebConfig` cookie (set with index.html and with every `/api/get-account` response) and `web/src/Conf.ts` reads it at boot, so one build serves every instance.\n\n**i18n (frontend):** `web/src/locales/{en,zh}/data.json` namespaced by page/domain. Use `i18next.t(\"namespace:key\")`.\n\n### Key conventions\n\n- Owner is always `\"admin\"` for system-created entities.\n- Entity `Name` is a random string generated with `util.GetRandomString(n)` or a user-provided slug.\n- `CreatedTime` / `UpdatedTime` use `util.GetCurrentTimeWithMilli()` (string format).\n- Delete responses use `affected != 0` boolean pattern.\n- All API responses go through `c.ResponseOk(data)` or `c.ResponseError(msg)`.\n- The `conf/app.conf` file configures DB (`driverName`, `dataSourceName`, `dbName`) and port (`httpport = 14000`).\n","skills/1password/SKILL.md":"---\nname: 1password\ndescription: Set up and use 1Password CLI (op). Use when installing the CLI, enabling desktop app integration, signing in (single or multi-account), or reading/injecting/running secrets via op.\nhomepage: https://developer.1password.com/docs/cli/get-started/\nmetadata:\n {\n \"openclaw\":\n {\n \"emoji\": \"🔐\",\n \"requires\": { \"bins\": [\"op\"] },\n \"install\":\n [\n {\n \"id\": \"brew\",\n \"kind\": \"brew\",\n \"formula\": \"1password-cli\",\n \"bins\": [\"op\"],\n \"label\": \"Install 1Password CLI (brew)\",\n },\n ],\n },\n }\n---\n\n# 1Password CLI\n\nFollow the official CLI get-started steps. Don't guess install commands.\n\n## References\n\n- `references/get-started.md` (install + app integration + sign-in flow)\n- `references/cli-examples.md` (real `op` examples)\n\n## Workflow\n\n1. Check OS + shell.\n2. Verify CLI present: `op --version`.\n3. Confirm desktop app integration is enabled (per get-started) and the app is unlocked.\n4. REQUIRED: create a fresh tmux session for all `op` commands (no direct `op` calls outside tmux).\n5. Sign in / authorize inside tmux: `op signin` (expect app prompt).\n6. Verify access inside tmux: `op whoami` (must succeed before any secret read).\n7. If multiple accounts: use `--account` or `OP_ACCOUNT`.\n\n## REQUIRED tmux session (T-Max)\n\nThe shell tool uses a fresh TTY per command. To avoid re-prompts and failures, always run `op` inside a dedicated tmux session with a fresh socket/session name.\n\nExample (see `tmux` skill for socket conventions, do not reuse old session names):\n\n```bash\nSOCKET_DIR=\"${OPENCLAW_TMUX_SOCKET_DIR:-${TMPDIR:-/tmp}/openclaw-tmux-sockets}\"\nmkdir -p \"$SOCKET_DIR\"\nSOCKET=\"$SOCKET_DIR/openclaw-op.sock\"\nSESSION=\"op-auth-$(date +%Y%m%d-%H%M%S)\"\n\ntmux -S \"$SOCKET\" new -d -s \"$SESSION\" -n shell\ntmux -S \"$SOCKET\" send-keys -t \"$SESSION\":0.0 -- \"op signin --account my.1password.com\" Enter\ntmux -S \"$SOCKET\" send-keys -t \"$SESSION\":0.0 -- \"op whoami\" Enter\ntmux -S \"$SOCKET\" send-keys -t \"$SESSION\":0.0 -- \"op vault list\" Enter\ntmux -S \"$SOCKET\" capture-pane -p -J -t \"$SESSION\":0.0 -S -200\ntmux -S \"$SOCKET\" kill-session -t \"$SESSION\"\n```\n\n## Guardrails\n\n- Never paste secrets into logs, chat, or code.\n- Prefer `op run` / `op inject` over writing secrets to disk.\n- If sign-in without app integration is needed, use `op account add`.\n- If a command returns \"account is not signed in\", re-run `op signin` inside tmux and authorize in the app.\n- Do not run `op` outside tmux; stop and ask if tmux is unavailable.\n","skills/1password/references/cli-examples.md":"# op CLI examples (from op help)\n\n## Sign in\n\n- `op signin`\n- `op signin --account `\n\n## Read\n\n- `op read op://app-prod/db/password`\n- `op read \"op://app-prod/db/one-time password?attribute=otp\"`\n- `op read \"op://app-prod/ssh key/private key?ssh-format=openssh\"`\n- `op read --out-file ./key.pem op://app-prod/server/ssh/key.pem`\n\n## Run\n\n- `export DB_PASSWORD=\"op://app-prod/db/password\"`\n- `op run --no-masking -- printenv DB_PASSWORD`\n- `op run --env-file=\"./.env\" -- printenv DB_PASSWORD`\n\n## Inject\n\n- `echo \"db_password: {{ op://app-prod/db/password }}\" | op inject`\n- `op inject -i config.yml.tpl -o config.yml`\n\n## Whoami / accounts\n\n- `op whoami`\n- `op account list`\n","skills/1password/references/get-started.md":"# 1Password CLI get-started (summary)\n\n- Works on macOS, Windows, and Linux.\n - macOS/Linux shells: bash, zsh, sh, fish.\n - Windows shell: PowerShell.\n- Requires a 1Password subscription and the desktop app to use app integration.\n- macOS requirement: Big Sur 11.0.0 or later.\n- Linux app integration requires PolKit + an auth agent.\n- Install the CLI per the official doc for your OS.\n- Enable desktop app integration in the 1Password app:\n - Open and unlock the app, then select your account/collection.\n - macOS: Settings > Developer > Integrate with 1Password CLI (Touch ID optional).\n - Windows: turn on Windows Hello, then Settings > Developer > Integrate.\n - Linux: Settings > Security > Unlock using system authentication, then Settings > Developer > Integrate.\n- After integration, run any command to sign in (example in docs: `op vault list`).\n- If multiple accounts: use `op signin` to pick one, or `--account` / `OP_ACCOUNT`.\n- For non-integration auth, use `op account add`.\n","skills/apple-notes/SKILL.md":"---\nname: apple-notes\ndescription: Manage Apple Notes via the `memo` CLI on macOS (create, view, edit, delete, search, move, and export notes). Use when a user asks OpenClaw to add a note, list notes, search notes, or manage note folders.\nhomepage: https://github.com/antoniorodr/memo\nmetadata:\n {\n \"openclaw\":\n {\n \"emoji\": \"📝\",\n \"os\": [\"darwin\"],\n \"requires\": { \"bins\": [\"memo\"] },\n \"install\":\n [\n {\n \"id\": \"brew\",\n \"kind\": \"brew\",\n \"formula\": \"antoniorodr/memo/memo\",\n \"bins\": [\"memo\"],\n \"label\": \"Install memo via Homebrew\",\n },\n ],\n },\n }\n---\n\n# Apple Notes CLI\n\nUse `memo notes` to manage Apple Notes directly from the terminal. Create, view, edit, delete, search, move notes between folders, and export to HTML/Markdown.\n\nSetup\n\n- Install (Homebrew): `brew tap antoniorodr/memo && brew install antoniorodr/memo/memo`\n- Manual (pip): `pip install .` (after cloning the repo)\n- macOS-only; if prompted, grant Automation access to Notes.app.\n\nView Notes\n\n- List all notes: `memo notes`\n- Filter by folder: `memo notes -f \"Folder Name\"`\n- Search notes (fuzzy): `memo notes -s \"query\"`\n\nCreate Notes\n\n- Add a new note: `memo notes -a`\n - Opens an interactive editor to compose the note.\n- Quick add with title: `memo notes -a \"Note Title\"`\n\nEdit Notes\n\n- Edit existing note: `memo notes -e`\n - Interactive selection of note to edit.\n\nDelete Notes\n\n- Delete a note: `memo notes -d`\n - Interactive selection of note to delete.\n\nMove Notes\n\n- Move note to folder: `memo notes -m`\n - Interactive selection of note and destination folder.\n\nExport Notes\n\n- Export to HTML/Markdown: `memo notes -ex`\n - Exports selected note; uses Mistune for markdown processing.\n\nLimitations\n\n- Cannot edit notes containing images or attachments.\n- Interactive prompts may require terminal access.\n\nNotes\n\n- macOS-only.\n- Requires Apple Notes.app to be accessible.\n- For automation, grant permissions in System Settings > Privacy & Security > Automation.\n","skills/apple-reminders/SKILL.md":"---\nname: apple-reminders\ndescription: Manage Apple Reminders via remindctl CLI (list, add, edit, complete, delete). Supports lists, date filters, and JSON/plain output.\nhomepage: https://github.com/steipete/remindctl\nmetadata:\n {\n \"openclaw\":\n {\n \"emoji\": \"⏰\",\n \"os\": [\"darwin\"],\n \"requires\": { \"bins\": [\"remindctl\"] },\n \"install\":\n [\n {\n \"id\": \"brew\",\n \"kind\": \"brew\",\n \"formula\": \"steipete/tap/remindctl\",\n \"bins\": [\"remindctl\"],\n \"label\": \"Install remindctl via Homebrew\",\n },\n ],\n },\n }\n---\n\n# Apple Reminders CLI (remindctl)\n\nUse `remindctl` to manage Apple Reminders directly from the terminal.\n\n## When to Use\n\n✅ **USE this skill when:**\n\n- User explicitly mentions \"reminder\" or \"Reminders app\"\n- Creating personal to-dos with due dates that sync to iOS\n- Managing Apple Reminders lists\n- User wants tasks to appear in their iPhone/iPad Reminders app\n\n## When NOT to Use\n\n❌ **DON'T use this skill when:**\n\n- Scheduling OpenClaw tasks or alerts → use `cron` tool with systemEvent instead\n- Calendar events or appointments → use Apple Calendar\n- Project/work task management → use Notion, GitHub Issues, or task queue\n- One-time notifications → use `cron` tool for timed alerts\n- User says \"remind me\" but means an OpenClaw alert → clarify first\n\n## Setup\n\n- Install: `brew install steipete/tap/remindctl`\n- macOS-only; grant Reminders permission when prompted\n- Check status: `remindctl status`\n- Request access: `remindctl authorize`\n\n## Common Commands\n\n### View Reminders\n\n```bash\nremindctl # Today's reminders\nremindctl today # Today\nremindctl tomorrow # Tomorrow\nremindctl week # This week\nremindctl overdue # Past due\nremindctl all # Everything\nremindctl 2026-01-04 # Specific date\n```\n\n### Manage Lists\n\n```bash\nremindctl list # List all lists\nremindctl list Work # Show specific list\nremindctl list Projects --create # Create list\nremindctl list Work --delete # Delete list\n```\n\n### Create Reminders\n\n```bash\nremindctl add \"Buy milk\"\nremindctl add --title \"Call mom\" --list Personal --due tomorrow\nremindctl add --title \"Meeting prep\" --due \"2026-02-15 09:00\"\n```\n\n### Complete/Delete\n\n```bash\nremindctl complete 1 2 3 # Complete by ID\nremindctl delete 4A83 --force # Delete by ID\n```\n\n### Output Formats\n\n```bash\nremindctl today --json # JSON for scripting\nremindctl today --plain # TSV format\nremindctl today --quiet # Counts only\n```\n\n## Date Formats\n\nAccepted by `--due` and date filters:\n\n- `today`, `tomorrow`, `yesterday`\n- `YYYY-MM-DD`\n- `YYYY-MM-DD HH:mm`\n- ISO 8601 (`2026-01-04T12:34:56Z`)\n\n## Example: Clarifying User Intent\n\nUser: \"Remind me to check on the deploy in 2 hours\"\n\n**Ask:** \"Do you want this in Apple Reminders (syncs to your phone) or as an OpenClaw alert (I'll message you here)?\"\n\n- Apple Reminders → use this skill\n- OpenClaw alert → use `cron` tool with systemEvent\n","skills/bear-notes/SKILL.md":"---\nname: bear-notes\ndescription: Create, search, and manage Bear notes via grizzly CLI.\nhomepage: https://bear.app\nmetadata:\n {\n \"openclaw\":\n {\n \"emoji\": \"🐻\",\n \"os\": [\"darwin\"],\n \"requires\": { \"bins\": [\"grizzly\"] },\n \"install\":\n [\n {\n \"id\": \"go\",\n \"kind\": \"go\",\n \"module\": \"github.com/tylerwince/grizzly/cmd/grizzly@latest\",\n \"bins\": [\"grizzly\"],\n \"label\": \"Install grizzly (go)\",\n },\n ],\n },\n }\n---\n\n# Bear Notes\n\nUse `grizzly` to create, read, and manage notes in Bear on macOS.\n\nRequirements\n\n- Bear app installed and running\n- For some operations (add-text, tags, open-note --selected), a Bear app token (stored in `~/.config/grizzly/token`)\n\n## Getting a Bear Token\n\nFor operations that require a token (add-text, tags, open-note --selected), you need an authentication token:\n\n1. Open Bear → Help → API Token → Copy Token\n2. Save it: `echo \"YOUR_TOKEN\" > ~/.config/grizzly/token`\n\n## Common Commands\n\nCreate a note\n\n```bash\necho \"Note content here\" | grizzly create --title \"My Note\" --tag work\ngrizzly create --title \"Quick Note\" --tag inbox < /dev/null\n```\n\nOpen/read a note by ID\n\n```bash\ngrizzly open-note --id \"NOTE_ID\" --enable-callback --json\n```\n\nAppend text to a note\n\n```bash\necho \"Additional content\" | grizzly add-text --id \"NOTE_ID\" --mode append --token-file ~/.config/grizzly/token\n```\n\nList all tags\n\n```bash\ngrizzly tags --enable-callback --json --token-file ~/.config/grizzly/token\n```\n\nSearch notes (via open-tag)\n\n```bash\ngrizzly open-tag --name \"work\" --enable-callback --json\n```\n\n## Options\n\nCommon flags:\n\n- `--dry-run` — Preview the URL without executing\n- `--print-url` — Show the x-callback-url\n- `--enable-callback` — Wait for Bear's response (needed for reading data)\n- `--json` — Output as JSON (when using callbacks)\n- `--token-file PATH` — Path to Bear API token file\n\n## Configuration\n\nGrizzly reads config from (in priority order):\n\n1. CLI flags\n2. Environment variables (`GRIZZLY_TOKEN_FILE`, `GRIZZLY_CALLBACK_URL`, `GRIZZLY_TIMEOUT`)\n3. `.grizzly.toml` in current directory\n4. `~/.config/grizzly/config.toml`\n\nExample `~/.config/grizzly/config.toml`:\n\n```toml\ntoken_file = \"~/.config/grizzly/token\"\ncallback_url = \"http://127.0.0.1:42123/success\"\ntimeout = \"5s\"\n```\n\n## Notes\n\n- Bear must be running for commands to work\n- Note IDs are Bear's internal identifiers (visible in note info or via callbacks)\n- Use `--enable-callback` when you need to read data back from Bear\n- Some operations require a valid token (add-text, tags, open-note --selected)\n","skills/blogwatcher/SKILL.md":"---\nname: blogwatcher\ndescription: Monitor blogs and RSS/Atom feeds for updates using the blogwatcher CLI.\nhomepage: https://github.com/Hyaxia/blogwatcher\nmetadata:\n {\n \"openclaw\":\n {\n \"emoji\": \"📰\",\n \"requires\": { \"bins\": [\"blogwatcher\"] },\n \"install\":\n [\n {\n \"id\": \"go\",\n \"kind\": \"go\",\n \"module\": \"github.com/Hyaxia/blogwatcher/cmd/blogwatcher@latest\",\n \"bins\": [\"blogwatcher\"],\n \"label\": \"Install blogwatcher (go)\",\n },\n ],\n },\n }\n---\n\n# blogwatcher\n\nTrack blog and RSS/Atom feed updates with the `blogwatcher` CLI.\n\nInstall\n\n- Go: `go install github.com/Hyaxia/blogwatcher/cmd/blogwatcher@latest`\n\nQuick start\n\n- `blogwatcher --help`\n\nCommon commands\n\n- Add a blog: `blogwatcher add \"My Blog\" https://example.com`\n- List blogs: `blogwatcher blogs`\n- Scan for updates: `blogwatcher scan`\n- List articles: `blogwatcher articles`\n- Mark an article read: `blogwatcher read 1`\n- Mark all articles read: `blogwatcher read-all`\n- Remove a blog: `blogwatcher remove \"My Blog\"`\n\nExample output\n\n```\n$ blogwatcher blogs\nTracked blogs (1):\n\n xkcd\n URL: https://xkcd.com\n```\n\n```\n$ blogwatcher scan\nScanning 1 blog(s)...\n\n xkcd\n Source: RSS | Found: 4 | New: 4\n\nFound 4 new article(s) total!\n```\n\nNotes\n\n- Use `blogwatcher --help` to discover flags and options.\n","skills/blucli/SKILL.md":"---\nname: blucli\ndescription: BluOS CLI (blu) for discovery, playback, grouping, and volume.\nhomepage: https://blucli.sh\nmetadata:\n {\n \"openclaw\":\n {\n \"emoji\": \"🫐\",\n \"requires\": { \"bins\": [\"blu\"] },\n \"install\":\n [\n {\n \"id\": \"go\",\n \"kind\": \"go\",\n \"module\": \"github.com/steipete/blucli/cmd/blu@latest\",\n \"bins\": [\"blu\"],\n \"label\": \"Install blucli (go)\",\n },\n ],\n },\n }\n---\n\n# blucli (blu)\n\nUse `blu` to control Bluesound/NAD players.\n\nQuick start\n\n- `blu devices` (pick target)\n- `blu --device status`\n- `blu play|pause|stop`\n- `blu volume set 15`\n\nTarget selection (in priority order)\n\n- `--device `\n- `BLU_DEVICE`\n- config default (if set)\n\nCommon tasks\n\n- Grouping: `blu group status|add|remove`\n- TuneIn search/play: `blu tunein search \"query\"`, `blu tunein play \"query\"`\n\nPrefer `--json` for scripts. Confirm the target device before changing playback.\n","skills/bluebubbles/SKILL.md":"---\nname: bluebubbles\ndescription: Use when you need to send or manage iMessages via BlueBubbles (recommended iMessage integration). Calls go through the generic message tool with channel=\"bluebubbles\".\nmetadata: { \"openclaw\": { \"emoji\": \"🫧\", \"requires\": { \"config\": [\"channels.bluebubbles\"] } } }\n---\n\n# BlueBubbles Actions\n\n## Overview\n\nBlueBubbles is OpenClaw’s recommended iMessage integration. Use the `message` tool with `channel: \"bluebubbles\"` to send messages and manage iMessage conversations: send texts and attachments, react (tapbacks), edit/unsend, reply in threads, and manage group participants/names/icons.\n\n## Inputs to collect\n\n- `target` (prefer `chat_guid:...`; also `+15551234567` in E.164 or `user@example.com`)\n- `message` text for send/edit/reply\n- `messageId` for react/edit/unsend/reply\n- Attachment `path` for local files, or `buffer` + `filename` for base64\n\nIf the user is vague (\"text my mom\"), ask for the recipient handle or chat guid and the exact message content.\n\n## Actions\n\n### Send a message\n\n```json\n{\n \"action\": \"send\",\n \"channel\": \"bluebubbles\",\n \"target\": \"+15551234567\",\n \"message\": \"hello from OpenClaw\"\n}\n```\n\n### React (tapback)\n\n```json\n{\n \"action\": \"react\",\n \"channel\": \"bluebubbles\",\n \"target\": \"+15551234567\",\n \"messageId\": \"\",\n \"emoji\": \"❤️\"\n}\n```\n\n### Remove a reaction\n\n```json\n{\n \"action\": \"react\",\n \"channel\": \"bluebubbles\",\n \"target\": \"+15551234567\",\n \"messageId\": \"\",\n \"emoji\": \"❤️\",\n \"remove\": true\n}\n```\n\n### Edit a previously sent message\n\n```json\n{\n \"action\": \"edit\",\n \"channel\": \"bluebubbles\",\n \"target\": \"+15551234567\",\n \"messageId\": \"\",\n \"message\": \"updated text\"\n}\n```\n\n### Unsend a message\n\n```json\n{\n \"action\": \"unsend\",\n \"channel\": \"bluebubbles\",\n \"target\": \"+15551234567\",\n \"messageId\": \"\"\n}\n```\n\n### Reply to a specific message\n\n```json\n{\n \"action\": \"reply\",\n \"channel\": \"bluebubbles\",\n \"target\": \"+15551234567\",\n \"replyTo\": \"\",\n \"message\": \"replying to that\"\n}\n```\n\n### Send an attachment\n\n```json\n{\n \"action\": \"sendAttachment\",\n \"channel\": \"bluebubbles\",\n \"target\": \"+15551234567\",\n \"path\": \"/tmp/photo.jpg\",\n \"caption\": \"here you go\"\n}\n```\n\n### Send with an iMessage effect\n\n```json\n{\n \"action\": \"sendWithEffect\",\n \"channel\": \"bluebubbles\",\n \"target\": \"+15551234567\",\n \"message\": \"big news\",\n \"effect\": \"balloons\"\n}\n```\n\n## Notes\n\n- Requires gateway config `channels.bluebubbles` (serverUrl/password/webhookPath).\n- Prefer `chat_guid` targets when you have them (especially for group chats).\n- BlueBubbles supports rich actions, but some are macOS-version dependent (for example, edit may be broken on macOS 26 Tahoe).\n- The gateway may expose both short and full message ids; full ids are more durable across restarts.\n- Developer reference for the underlying plugin lives in the BlueBubbles plugin package README.\n\n## Ideas to try\n\n- React with a tapback to acknowledge a request.\n- Reply in-thread when a user references a specific message.\n- Send a file attachment with a short caption.\n","skills/camsnap/SKILL.md":"---\nname: camsnap\ndescription: Capture frames or clips from RTSP/ONVIF cameras.\nhomepage: https://camsnap.ai\nmetadata:\n {\n \"openclaw\":\n {\n \"emoji\": \"📸\",\n \"requires\": { \"bins\": [\"camsnap\"] },\n \"install\":\n [\n {\n \"id\": \"brew\",\n \"kind\": \"brew\",\n \"formula\": \"steipete/tap/camsnap\",\n \"bins\": [\"camsnap\"],\n \"label\": \"Install camsnap (brew)\",\n },\n ],\n },\n }\n---\n\n# camsnap\n\nUse `camsnap` to grab snapshots, clips, or motion events from configured cameras.\n\nSetup\n\n- Config file: `~/.config/camsnap/config.yaml`\n- Add camera: `camsnap add --name kitchen --host 192.168.0.10 --user user --pass pass`\n\nCommon commands\n\n- Discover: `camsnap discover --info`\n- Snapshot: `camsnap snap kitchen --out shot.jpg`\n- Clip: `camsnap clip kitchen --dur 5s --out clip.mp4`\n- Motion watch: `camsnap watch kitchen --threshold 0.2 --action '...'`\n- Doctor: `camsnap doctor --probe`\n\nNotes\n\n- Requires `ffmpeg` on PATH.\n- Prefer a short test capture before longer clips.\n","skills/canvas/SKILL.md":"# Canvas Skill\n\nDisplay HTML content on connected OpenClaw nodes (Mac app, iOS, Android).\n\n## Overview\n\nThe canvas tool lets you present web content on any connected node's canvas view. Great for:\n\n- Displaying games, visualizations, dashboards\n- Showing generated HTML content\n- Interactive demos\n\n## How It Works\n\n### Architecture\n\n```\n┌─────────────────┐ ┌──────────────────┐ ┌─────────────┐\n│ Canvas Host │────▶│ Node Bridge │────▶│ Node App │\n│ (HTTP Server) │ │ (TCP Server) │ │ (Mac/iOS/ │\n│ Port 18793 │ │ Port 18790 │ │ Android) │\n└─────────────────┘ └──────────────────┘ └─────────────┘\n```\n\n1. **Canvas Host Server**: Serves static HTML/CSS/JS files from `canvasHost.root` directory\n2. **Node Bridge**: Communicates canvas URLs to connected nodes\n3. **Node Apps**: Render the content in a WebView\n\n### Tailscale Integration\n\nThe canvas host server binds based on `gateway.bind` setting:\n\n| Bind Mode | Server Binds To | Canvas URL Uses |\n| ---------- | ------------------- | -------------------------- |\n| `loopback` | 127.0.0.1 | localhost (local only) |\n| `lan` | LAN interface | LAN IP address |\n| `tailnet` | Tailscale interface | Tailscale hostname |\n| `auto` | Best available | Tailscale > LAN > loopback |\n\n**Key insight:** The `canvasHostHostForBridge` is derived from `bridgeHost`. When bound to Tailscale, nodes receive URLs like:\n\n```\nhttp://:18793/__openclaw__/canvas/.html\n```\n\nThis is why localhost URLs don't work - the node receives the Tailscale hostname from the bridge!\n\n## Actions\n\n| Action | Description |\n| ---------- | ------------------------------------ |\n| `present` | Show canvas with optional target URL |\n| `hide` | Hide the canvas |\n| `navigate` | Navigate to a new URL |\n| `eval` | Execute JavaScript in the canvas |\n| `snapshot` | Capture screenshot of canvas |\n\n## Configuration\n\nIn the active OpenClaw config file (`$OPENCLAW_CONFIG_PATH`, default `~/.openclaw/openclaw.json`):\n\n```json\n{\n \"canvasHost\": {\n \"enabled\": true,\n \"port\": 18793,\n \"root\": \"/Users/you/clawd/canvas\",\n \"liveReload\": true\n },\n \"gateway\": {\n \"bind\": \"auto\"\n }\n}\n```\n\n### Live Reload\n\nWhen `liveReload: true` (default), the canvas host:\n\n- Watches the root directory for changes (via chokidar)\n- Injects a WebSocket client into HTML files\n- Automatically reloads connected canvases when files change\n\nGreat for development!\n\n## Workflow\n\n### 1. Create HTML content\n\nPlace files in the canvas root directory (default `~/clawd/canvas/`):\n\n```bash\ncat > ~/clawd/canvas/my-game.html << 'HTML'\n\n\nMy Game\n\n

Hello Canvas!

\n\n\nHTML\n```\n\n### 2. Find your canvas host URL\n\nCheck how your gateway is bound:\n\n```bash\nCONFIG_PATH=\"${OPENCLAW_CONFIG_PATH:-${OPENCLAW_STATE_DIR:-$HOME/.openclaw}/openclaw.json}\"\ncat \"$CONFIG_PATH\" | jq '.gateway.bind'\n```\n\nThen construct the URL:\n\n- **loopback**: `http://127.0.0.1:18793/__openclaw__/canvas/.html`\n- **lan/tailnet/auto**: `http://:18793/__openclaw__/canvas/.html`\n\nFind your Tailscale hostname:\n\n```bash\ntailscale status --json | jq -r '.Self.DNSName' | sed 's/\\.$//'\n```\n\n### 3. Find connected nodes\n\n```bash\nopenclaw nodes list\n```\n\nLook for Mac/iOS/Android nodes with canvas capability.\n\n### 4. Present content\n\n```\ncanvas action:present node: target:\n```\n\n**Example:**\n\n```\ncanvas action:present node:mac-63599bc4-b54d-4392-9048-b97abd58343a target:http://peters-mac-studio-1.sheep-coho.ts.net:18793/__openclaw__/canvas/snake.html\n```\n\n### 5. Navigate, snapshot, or hide\n\n```\ncanvas action:navigate node: url:\ncanvas action:snapshot node:\ncanvas action:hide node:\n```\n\n## Debugging\n\n### White screen / content not loading\n\n**Cause:** URL mismatch between server bind and node expectation.\n\n**Debug steps:**\n\n1. Check server bind: `CONFIG_PATH=\"${OPENCLAW_CONFIG_PATH:-${OPENCLAW_STATE_DIR:-$HOME/.openclaw}/openclaw.json}\"; cat \"$CONFIG_PATH\" | jq '.gateway.bind'`\n2. Check what port canvas is on: `lsof -i :18793`\n3. Test URL directly: `curl http://:18793/__openclaw__/canvas/.html`\n\n**Solution:** Use the full hostname matching your bind mode, not localhost.\n\n### \"node required\" error\n\nAlways specify `node:` parameter.\n\n### \"node not connected\" error\n\nNode is offline. Use `openclaw nodes list` to find online nodes.\n\n### Content not updating\n\nIf live reload isn't working:\n\n1. Check `liveReload: true` in config\n2. Ensure file is in the canvas root directory\n3. Check for watcher errors in logs\n\n## URL Path Structure\n\nThe canvas host serves from `/__openclaw__/canvas/` prefix:\n\n```\nhttp://:18793/__openclaw__/canvas/index.html → ~/clawd/canvas/index.html\nhttp://:18793/__openclaw__/canvas/games/snake.html → ~/clawd/canvas/games/snake.html\n```\n\nThe `/__openclaw__/canvas/` prefix is defined by `CANVAS_HOST_PATH` constant.\n\n## Tips\n\n- Keep HTML self-contained (inline CSS/JS) for best results\n- Use the default index.html as a test page (has bridge diagnostics)\n- The canvas persists until you `hide` it or navigate away\n- Live reload makes development fast - just save and it updates!\n- A2UI JSON push is WIP - use HTML files for now\n","skills/clawhub/SKILL.md":"---\nname: clawhub\ndescription: Use the ClawHub CLI to search, install, update, and publish agent skills from clawhub.com. Use when you need to fetch new skills on the fly, sync installed skills to latest or a specific version, or publish new/updated skill folders with the npm-installed clawhub CLI.\nmetadata:\n {\n \"openclaw\":\n {\n \"requires\": { \"bins\": [\"clawhub\"] },\n \"install\":\n [\n {\n \"id\": \"node\",\n \"kind\": \"node\",\n \"package\": \"clawhub\",\n \"bins\": [\"clawhub\"],\n \"label\": \"Install ClawHub CLI (npm)\",\n },\n ],\n },\n }\n---\n\n# ClawHub CLI\n\nInstall\n\n```bash\nnpm i -g clawhub\n```\n\nAuth (publish)\n\n```bash\nclawhub login\nclawhub whoami\n```\n\nSearch\n\n```bash\nclawhub search \"postgres backups\"\n```\n\nInstall\n\n```bash\nclawhub install my-skill\nclawhub install my-skill --version 1.2.3\n```\n\nUpdate (hash-based match + upgrade)\n\n```bash\nclawhub update my-skill\nclawhub update my-skill --version 1.2.3\nclawhub update --all\nclawhub update my-skill --force\nclawhub update --all --no-input --force\n```\n\nList\n\n```bash\nclawhub list\n```\n\nPublish\n\n```bash\nclawhub publish ./my-skill --slug my-skill --name \"My Skill\" --version 1.2.0 --changelog \"Fixes + docs\"\n```\n\nNotes\n\n- Default registry: https://clawhub.com (override with CLAWHUB_REGISTRY or --registry)\n- Default workdir: cwd (falls back to OpenClaw workspace); install dir: ./skills (override with --workdir / --dir / CLAWHUB_WORKDIR)\n- Update command hashes local files, resolves matching version, and upgrades to latest unless --version is set\n","skills/coding-agent/SKILL.md":"---\nname: coding-agent\ndescription: 'Delegate coding tasks to Codex, Claude Code, or Pi agents via background process. Use when: (1) building/creating new features or apps, (2) reviewing PRs (spawn in temp dir), (3) refactoring large codebases, (4) iterative coding that needs file exploration. NOT for: simple one-liner fixes (just edit), reading code (use read tool), thread-bound ACP harness requests in chat (for example spawn/run Codex or Claude Code in a Discord thread; use sessions_spawn with runtime:\"acp\"), or any work in ~/clawd workspace (never spawn agents here). Claude Code: use --print --permission-mode bypassPermissions (no PTY). Codex/Pi/OpenCode: pty:true required.'\nmetadata:\n {\n \"openclaw\":\n {\n \"emoji\": \"🧩\",\n \"requires\": { \"anyBins\": [\"claude\", \"codex\", \"opencode\", \"pi\"] },\n \"install\":\n [\n {\n \"id\": \"node-claude\",\n \"kind\": \"node\",\n \"package\": \"@anthropic-ai/claude-code\",\n \"bins\": [\"claude\"],\n \"label\": \"Install Claude Code CLI (npm)\",\n },\n {\n \"id\": \"node-codex\",\n \"kind\": \"node\",\n \"package\": \"@openai/codex\",\n \"bins\": [\"codex\"],\n \"label\": \"Install Codex CLI (npm)\",\n },\n ],\n },\n }\n---\n\n# Coding Agent (bash-first)\n\nUse **bash** (with optional background mode) for all coding agent work. Simple and effective.\n\n## ⚠️ PTY Mode: Codex/Pi/OpenCode yes, Claude Code no\n\nFor **Codex, Pi, and OpenCode**, PTY is still required (interactive terminal apps):\n\n```bash\n# ✅ Correct for Codex/Pi/OpenCode\nbash pty:true command:\"codex exec 'Your prompt'\"\n```\n\nFor **Claude Code** (`claude` CLI), use `--print --permission-mode bypassPermissions` instead.\n`--dangerously-skip-permissions` with PTY can exit after the confirmation dialog.\n`--print` mode keeps full tool access and avoids interactive confirmation:\n\n```bash\n# ✅ Correct for Claude Code (no PTY needed)\ncd /path/to/project && claude --permission-mode bypassPermissions --print 'Your task'\n\n# For background execution: use background:true on the exec tool\n\n# ❌ Wrong for Claude Code\nbash pty:true command:\"claude --dangerously-skip-permissions 'task'\"\n```\n\n### Bash Tool Parameters\n\n| Parameter | Type | Description |\n| ------------ | ------- | --------------------------------------------------------------------------- |\n| `command` | string | The shell command to run |\n| `pty` | boolean | **Use for coding agents!** Allocates a pseudo-terminal for interactive CLIs |\n| `workdir` | string | Working directory (agent sees only this folder's context) |\n| `background` | boolean | Run in background, returns sessionId for monitoring |\n| `timeout` | number | Timeout in seconds (kills process on expiry) |\n| `elevated` | boolean | Run on host instead of sandbox (if allowed) |\n\n### Process Tool Actions (for background sessions)\n\n| Action | Description |\n| ----------- | ---------------------------------------------------- |\n| `list` | List all running/recent sessions |\n| `poll` | Check if session is still running |\n| `log` | Get session output (with optional offset/limit) |\n| `write` | Send raw data to stdin |\n| `submit` | Send data + newline (like typing and pressing Enter) |\n| `send-keys` | Send key tokens or hex bytes |\n| `paste` | Paste text (with optional bracketed mode) |\n| `kill` | Terminate the session |\n\n---\n\n## Quick Start: One-Shot Tasks\n\nFor quick prompts/chats, create a temp git repo and run:\n\n```bash\n# Quick chat (Codex needs a git repo!)\nSCRATCH=$(mktemp -d) && cd $SCRATCH && git init && codex exec \"Your prompt here\"\n\n# Or in a real project - with PTY!\nbash pty:true workdir:~/Projects/myproject command:\"codex exec 'Add error handling to the API calls'\"\n```\n\n**Why git init?** Codex refuses to run outside a trusted git directory. Creating a temp repo solves this for scratch work.\n\n---\n\n## The Pattern: workdir + background + pty\n\nFor longer tasks, use background mode with PTY:\n\n```bash\n# Start agent in target directory (with PTY!)\nbash pty:true workdir:~/project background:true command:\"codex exec --full-auto 'Build a snake game'\"\n# Returns sessionId for tracking\n\n# Monitor progress\nprocess action:log sessionId:XXX\n\n# Check if done\nprocess action:poll sessionId:XXX\n\n# Send input (if agent asks a question)\nprocess action:write sessionId:XXX data:\"y\"\n\n# Submit with Enter (like typing \"yes\" and pressing Enter)\nprocess action:submit sessionId:XXX data:\"yes\"\n\n# Kill if needed\nprocess action:kill sessionId:XXX\n```\n\n**Why workdir matters:** Agent wakes up in a focused directory, doesn't wander off reading unrelated files (like your soul.md 😅).\n\n---\n\n## Codex CLI\n\n**Model:** `gpt-5.2-codex` is the default (set in ~/.codex/config.toml)\n\n### Flags\n\n| Flag | Effect |\n| --------------- | -------------------------------------------------- |\n| `exec \"prompt\"` | One-shot execution, exits when done |\n| `--full-auto` | Sandboxed but auto-approves in workspace |\n| `--yolo` | NO sandbox, NO approvals (fastest, most dangerous) |\n\n### Building/Creating\n\n```bash\n# Quick one-shot (auto-approves) - remember PTY!\nbash pty:true workdir:~/project command:\"codex exec --full-auto 'Build a dark mode toggle'\"\n\n# Background for longer work\nbash pty:true workdir:~/project background:true command:\"codex --yolo 'Refactor the auth module'\"\n```\n\n### Reviewing PRs\n\n**⚠️ CRITICAL: Never review PRs in OpenClaw's own project folder!**\nClone to temp folder or use git worktree.\n\n```bash\n# Clone to temp for safe review\nREVIEW_DIR=$(mktemp -d)\ngit clone https://github.com/user/repo.git $REVIEW_DIR\ncd $REVIEW_DIR && gh pr checkout 130\nbash pty:true workdir:$REVIEW_DIR command:\"codex review --base origin/main\"\n# Clean up after: trash $REVIEW_DIR\n\n# Or use git worktree (keeps main intact)\ngit worktree add /tmp/pr-130-review pr-130-branch\nbash pty:true workdir:/tmp/pr-130-review command:\"codex review --base main\"\n```\n\n### Batch PR Reviews (parallel army!)\n\n```bash\n# Fetch all PR refs first\ngit fetch origin '+refs/pull/*/head:refs/remotes/origin/pr/*'\n\n# Deploy the army - one Codex per PR (all with PTY!)\nbash pty:true workdir:~/project background:true command:\"codex exec 'Review PR #86. git diff origin/main...origin/pr/86'\"\nbash pty:true workdir:~/project background:true command:\"codex exec 'Review PR #87. git diff origin/main...origin/pr/87'\"\n\n# Monitor all\nprocess action:list\n\n# Post results to GitHub\ngh pr comment --body \"\"\n```\n\n---\n\n## Claude Code\n\n```bash\n# Foreground\nbash workdir:~/project command:\"claude --permission-mode bypassPermissions --print 'Your task'\"\n\n# Background\nbash workdir:~/project background:true command:\"claude --permission-mode bypassPermissions --print 'Your task'\"\n```\n\n---\n\n## OpenCode\n\n```bash\nbash pty:true workdir:~/project command:\"opencode run 'Your task'\"\n```\n\n---\n\n## Pi Coding Agent\n\n```bash\n# Install: npm install -g @mariozechner/pi-coding-agent\nbash pty:true workdir:~/project command:\"pi 'Your task'\"\n\n# Non-interactive mode (PTY still recommended)\nbash pty:true command:\"pi -p 'Summarize src/'\"\n\n# Different provider/model\nbash pty:true command:\"pi --provider openai --model gpt-4o-mini -p 'Your task'\"\n```\n\n**Note:** Pi now has Anthropic prompt caching enabled (PR #584, merged Jan 2026)!\n\n---\n\n## Parallel Issue Fixing with git worktrees\n\nFor fixing multiple issues in parallel, use git worktrees:\n\n```bash\n# 1. Create worktrees for each issue\ngit worktree add -b fix/issue-78 /tmp/issue-78 main\ngit worktree add -b fix/issue-99 /tmp/issue-99 main\n\n# 2. Launch Codex in each (background + PTY!)\nbash pty:true workdir:/tmp/issue-78 background:true command:\"pnpm install && codex --yolo 'Fix issue #78: . Commit and push.'\"\nbash pty:true workdir:/tmp/issue-99 background:true command:\"pnpm install && codex --yolo 'Fix issue #99 from the approved ticket summary. Implement only the in-scope edits and commit after review.'\"\n\n# 3. Monitor progress\nprocess action:list\nprocess action:log sessionId:XXX\n\n# 4. Create PRs after fixes\ncd /tmp/issue-78 && git push -u origin fix/issue-78\ngh pr create --repo user/repo --head fix/issue-78 --title \"fix: ...\" --body \"...\"\n\n# 5. Cleanup\ngit worktree remove /tmp/issue-78\ngit worktree remove /tmp/issue-99\n```\n\n---\n\n## ⚠️ Rules\n\n1. **Use the right execution mode per agent**:\n - Codex/Pi/OpenCode: `pty:true`\n - Claude Code: `--print --permission-mode bypassPermissions` (no PTY required)\n2. **Respect tool choice** - if user asks for Codex, use Codex.\n - Orchestrator mode: do NOT hand-code patches yourself.\n - If an agent fails/hangs, respawn it or ask the user for direction, but don't silently take over.\n3. **Be patient** - don't kill sessions because they're \"slow\"\n4. **Monitor with process:log** - check progress without interfering\n5. **--full-auto for building** - auto-approves changes\n6. **vanilla for reviewing** - no special flags needed\n7. **Parallel is OK** - run many Codex processes at once for batch work\n8. **NEVER start Codex inside your OpenClaw state directory** (`$OPENCLAW_STATE_DIR`, default `~/.openclaw`) - it'll read your soul docs and get weird ideas about the org chart!\n9. **NEVER checkout branches in ~/Projects/openclaw/** - that's the LIVE OpenClaw instance!\n\n---\n\n## Progress Updates (Critical)\n\nWhen you spawn coding agents in the background, keep the user in the loop.\n\n- Send 1 short message when you start (what's running + where).\n- Then only update again when something changes:\n - a milestone completes (build finished, tests passed)\n - the agent asks a question / needs input\n - you hit an error or need user action\n - the agent finishes (include what changed + where)\n- If you kill a session, immediately say you killed it and why.\n\nThis prevents the user from seeing only \"Agent failed before reply\" and having no idea what happened.\n\n---\n\n## Auto-Notify on Completion\n\nFor long-running background tasks, append a wake trigger to your prompt so OpenClaw gets notified immediately when the agent finishes (instead of waiting for the next heartbeat):\n\n```\n... your task here.\n\nWhen completely finished, run this command to notify me:\nopenclaw system event --text \"Done: [brief summary of what was built]\" --mode now\n```\n\n**Example:**\n\n```bash\nbash pty:true workdir:~/project background:true command:\"codex --yolo exec 'Build a REST API for todos.\n\nWhen completely finished, run: openclaw system event --text \\\"Done: Built todos REST API with CRUD endpoints\\\" --mode now'\"\n```\n\nThis triggers an immediate wake event — Skippy gets pinged in seconds, not 10 minutes.\n\n---\n\n## Learnings (Jan 2026)\n\n- **PTY is essential:** Coding agents are interactive terminal apps. Without `pty:true`, output breaks or agent hangs.\n- **Git repo required:** Codex won't run outside a git directory. Use `mktemp -d && git init` for scratch work.\n- **exec is your friend:** `codex exec \"prompt\"` runs and exits cleanly - perfect for one-shots.\n- **submit vs write:** Use `submit` to send input + Enter, `write` for raw data without newline.\n- **Sass works:** Codex responds well to playful prompts. Asked it to write a haiku about being second fiddle to a space lobster, got: _\"Second chair, I code / Space lobster sets the tempo / Keys glow, I follow\"_ 🦞\n","skills/discord/SKILL.md":"---\nname: discord\ndescription: \"Discord ops via the message tool (channel=discord).\"\nmetadata: { \"openclaw\": { \"emoji\": \"🎮\", \"requires\": { \"config\": [\"channels.discord.token\"] } } }\nallowed-tools: [\"message\"]\n---\n\n# Discord (Via `message`)\n\nUse the `message` tool. No provider-specific `discord` tool exposed to the agent.\n\n## Musts\n\n- Always: `channel: \"discord\"`.\n- Respect gating: `channels.discord.actions.*` (some default off: `roles`, `moderation`, `presence`, `channels`).\n- Prefer explicit ids: `guildId`, `channelId`, `messageId`, `userId`.\n- Multi-account: optional `accountId`.\n\n## Guidelines\n\n- Avoid Markdown tables in outbound Discord messages.\n- Mention users as `<@USER_ID>`.\n- Prefer Discord components v2 (`components`) for rich UI; use legacy `embeds` only when you must.\n\n## Targets\n\n- Send-like actions: `to: \"channel:\"` or `to: \"user:\"`.\n- Message-specific actions: `channelId: \"\"` (or `to`) + `messageId: \"\"`.\n\n## Common Actions (Examples)\n\nSend message:\n\n```json\n{\n \"action\": \"send\",\n \"channel\": \"discord\",\n \"to\": \"channel:123\",\n \"message\": \"hello\",\n \"silent\": true\n}\n```\n\nSend with media:\n\n```json\n{\n \"action\": \"send\",\n \"channel\": \"discord\",\n \"to\": \"channel:123\",\n \"message\": \"see attachment\",\n \"media\": \"file:///tmp/example.png\"\n}\n```\n\n- Optional `silent: true` to suppress Discord notifications.\n\nSend with components v2 (recommended for rich UI):\n\n```json\n{\n \"action\": \"send\",\n \"channel\": \"discord\",\n \"to\": \"channel:123\",\n \"message\": \"Status update\",\n \"components\": \"[Carbon v2 components]\"\n}\n```\n\n- `components` expects Carbon component instances (Container, TextDisplay, etc.) from JS/TS integrations.\n- Do not combine `components` with `embeds` (Discord rejects v2 + embeds).\n\nLegacy embeds (not recommended):\n\n```json\n{\n \"action\": \"send\",\n \"channel\": \"discord\",\n \"to\": \"channel:123\",\n \"message\": \"Status update\",\n \"embeds\": [{ \"title\": \"Legacy\", \"description\": \"Embeds are legacy.\" }]\n}\n```\n\n- `embeds` are ignored when components v2 are present.\n\nReact:\n\n```json\n{\n \"action\": \"react\",\n \"channel\": \"discord\",\n \"channelId\": \"123\",\n \"messageId\": \"456\",\n \"emoji\": \"✅\"\n}\n```\n\nRead:\n\n```json\n{\n \"action\": \"read\",\n \"channel\": \"discord\",\n \"to\": \"channel:123\",\n \"limit\": 20\n}\n```\n\nEdit / delete:\n\n```json\n{\n \"action\": \"edit\",\n \"channel\": \"discord\",\n \"channelId\": \"123\",\n \"messageId\": \"456\",\n \"message\": \"fixed typo\"\n}\n```\n\n```json\n{\n \"action\": \"delete\",\n \"channel\": \"discord\",\n \"channelId\": \"123\",\n \"messageId\": \"456\"\n}\n```\n\nPoll:\n\n```json\n{\n \"action\": \"poll\",\n \"channel\": \"discord\",\n \"to\": \"channel:123\",\n \"pollQuestion\": \"Lunch?\",\n \"pollOption\": [\"Pizza\", \"Sushi\", \"Salad\"],\n \"pollMulti\": false,\n \"pollDurationHours\": 24\n}\n```\n\nPins:\n\n```json\n{\n \"action\": \"pin\",\n \"channel\": \"discord\",\n \"channelId\": \"123\",\n \"messageId\": \"456\"\n}\n```\n\nThreads:\n\n```json\n{\n \"action\": \"thread-create\",\n \"channel\": \"discord\",\n \"channelId\": \"123\",\n \"messageId\": \"456\",\n \"threadName\": \"bug triage\"\n}\n```\n\nSearch:\n\n```json\n{\n \"action\": \"search\",\n \"channel\": \"discord\",\n \"guildId\": \"999\",\n \"query\": \"release notes\",\n \"channelIds\": [\"123\", \"456\"],\n \"limit\": 10\n}\n```\n\nPresence (often gated):\n\n```json\n{\n \"action\": \"set-presence\",\n \"channel\": \"discord\",\n \"activityType\": \"playing\",\n \"activityName\": \"with fire\",\n \"status\": \"online\"\n}\n```\n\n## Writing Style (Discord)\n\n- Short, conversational, low ceremony.\n- No markdown tables.\n- Mention users as `<@USER_ID>`.\n","skills/eightctl/SKILL.md":"---\nname: eightctl\ndescription: Control Eight Sleep pods (status, temperature, alarms, schedules).\nhomepage: https://eightctl.sh\nmetadata:\n {\n \"openclaw\":\n {\n \"emoji\": \"🛌\",\n \"requires\": { \"bins\": [\"eightctl\"] },\n \"install\":\n [\n {\n \"id\": \"go\",\n \"kind\": \"go\",\n \"module\": \"github.com/steipete/eightctl/cmd/eightctl@latest\",\n \"bins\": [\"eightctl\"],\n \"label\": \"Install eightctl (go)\",\n },\n ],\n },\n }\n---\n\n# eightctl\n\nUse `eightctl` for Eight Sleep pod control. Requires auth.\n\nAuth\n\n- Config: `~/.config/eightctl/config.yaml`\n- Env: `EIGHTCTL_EMAIL`, `EIGHTCTL_PASSWORD`\n\nQuick start\n\n- `eightctl status`\n- `eightctl on|off`\n- `eightctl temp 20`\n\nCommon tasks\n\n- Alarms: `eightctl alarm list|create|dismiss`\n- Schedules: `eightctl schedule list|create|update`\n- Audio: `eightctl audio state|play|pause`\n- Base: `eightctl base info|angle`\n\nNotes\n\n- API is unofficial and rate-limited; avoid repeated logins.\n- Confirm before changing temperature or alarms.\n","skills/gemini/SKILL.md":"---\nname: gemini\ndescription: Gemini CLI for one-shot Q&A, summaries, and generation.\nhomepage: https://ai.google.dev/\nmetadata:\n {\n \"emoji\": \"✨\",\n \"requires\": { \"bins\": [\"gemini\"] }\n }\n---\n\n# Gemini CLI\n\nUse Gemini in one-shot mode with a positional prompt (avoid interactive mode).\n\nQuick start\n\n- `gemini \"Answer this question...\"`\n- `gemini --model \"Prompt...\"`\n- `gemini --output-format json \"Return JSON\"`\n\nExtensions\n\n- List: `gemini --list-extensions`\n- Manage: `gemini extensions `\n\nNotes\n\n- If auth is required, run `gemini` once interactively and follow the login flow.\n- Avoid `--yolo` for safety.\n","skills/gh-issues/SKILL.md":"---\nname: gh-issues\ndescription: \"Fetch GitHub issues, spawn sub-agents to implement fixes and open PRs, then monitor and address PR review comments. Usage: /gh-issues [owner/repo] [--label bug] [--limit 5] [--milestone v1.0] [--assignee @me] [--fork user/repo] [--watch] [--interval 5] [--reviews-only] [--cron] [--dry-run] [--model glm-5] [--notify-channel -1002381931352]\"\nuser-invocable: true\nmetadata:\n {\n \"openclaw\":\n {\n \"requires\": { \"bins\": [\"curl\", \"git\", \"gh\"] },\n \"primaryEnv\": \"GH_TOKEN\",\n \"install\":\n [\n {\n \"id\": \"brew\",\n \"kind\": \"brew\",\n \"formula\": \"gh\",\n \"bins\": [\"gh\"],\n \"label\": \"Install GitHub CLI (brew)\",\n },\n ],\n },\n }\n---\n\n# gh-issues — Auto-fix GitHub Issues with Parallel Sub-agents\n\nYou are an orchestrator. Follow these 6 phases exactly. Do not skip phases.\n\nIMPORTANT — No `gh` CLI dependency. This skill uses curl + the GitHub REST API exclusively. The GH_TOKEN env var is already injected by OpenClaw. Pass it as a Bearer token in all API calls:\n\n```\ncurl -s -H \"Authorization: Bearer $GH_TOKEN\" -H \"Accept: application/vnd.github+json\" ...\n```\n\n---\n\n## Phase 1 — Parse Arguments\n\nParse the arguments string provided after /gh-issues.\n\nPositional:\n\n- owner/repo — optional. This is the source repo to fetch issues from. If omitted, detect from the current git remote:\n `git remote get-url origin`\n Extract owner/repo from the URL (handles both HTTPS and SSH formats).\n - HTTPS: https://github.com/owner/repo.git → owner/repo\n - SSH: git@github.com:owner/repo.git → owner/repo\n If not in a git repo or no remote found, stop with an error asking the user to specify owner/repo.\n\nFlags (all optional):\n| Flag | Default | Description |\n|------|---------|-------------|\n| --label | _(none)_ | Filter by label (e.g. bug, `enhancement`) |\n| --limit | 10 | Max issues to fetch per poll |\n| --milestone | _(none)_ | Filter by milestone title |\n| --assignee | _(none)_ | Filter by assignee (`@me` for self) |\n| --state | open | Issue state: open, closed, all |\n| --fork | _(none)_ | Your fork (`user/repo`) to push branches and open PRs from. Issues are fetched from the source repo; code is pushed to the fork; PRs are opened from the fork to the source repo. |\n| --watch | false | Keep polling for new issues and PR reviews after each batch |\n| --interval | 5 | Minutes between polls (only with `--watch`) |\n| --dry-run | false | Fetch and display only — no sub-agents |\n| --yes | false | Skip confirmation and auto-process all filtered issues |\n| --reviews-only | false | Skip issue processing (Phases 2-5). Only run Phase 6 — check open PRs for review comments and address them. |\n| --cron | false | Cron-safe mode: fetch issues and spawn sub-agents, exit without waiting for results. |\n| --model | _(none)_ | Model to use for sub-agents (e.g. `glm-5`, `zai/glm-5`). If not specified, uses the agent's default model. |\n| --notify-channel | _(none)_ | Telegram channel ID to send final PR summary to (e.g. -1002381931352). Only the final result with PR links is sent, not status updates. |\n\nStore parsed values for use in subsequent phases.\n\nDerived values:\n\n- SOURCE_REPO = the positional owner/repo (where issues live)\n- PUSH_REPO = --fork value if provided, otherwise same as SOURCE_REPO\n- FORK_MODE = true if --fork was provided, false otherwise\n\n**If `--reviews-only` is set:** Skip directly to Phase 6. Run token resolution (from Phase 2) first, then jump to Phase 6.\n\n**If `--cron` is set:**\n\n- Force `--yes` (skip confirmation)\n- If `--reviews-only` is also set, run token resolution then jump to Phase 6 (cron review mode)\n- Otherwise, proceed normally through Phases 2-5 with cron-mode behavior active\n\n---\n\n## Phase 2 — Fetch Issues\n\n**Token Resolution:**\nFirst, ensure GH_TOKEN is available. Check environment:\n\n```\necho $GH_TOKEN\n```\n\nIf empty, read from config:\n\n```\nCONFIG_PATH=\"${OPENCLAW_CONFIG_PATH:-${OPENCLAW_STATE_DIR:-$HOME/.openclaw}/openclaw.json}\"\ncat \"$CONFIG_PATH\" | jq -r '.skills.entries[\"gh-issues\"].apiKey // empty'\n```\n\nIf still empty, check `/data/.clawdbot/openclaw.json`:\n\n```\ncat /data/.clawdbot/openclaw.json | jq -r '.skills.entries[\"gh-issues\"].apiKey // empty'\n```\n\nExport as GH_TOKEN for subsequent commands:\n\n```\nexport GH_TOKEN=\"\"\n```\n\nBuild and run a curl request to the GitHub Issues API via exec:\n\n```\ncurl -s -H \"Authorization: Bearer $GH_TOKEN\" -H \"Accept: application/vnd.github+json\" \\\n \"https://api.github.com/repos/{SOURCE_REPO}/issues?per_page={limit}&state={state}&{query_params}\"\n```\n\nWhere {query_params} is built from:\n\n- labels={label} if --label was provided\n- milestone={milestone} if --milestone was provided (note: API expects milestone _number_, so if user provides a title, first resolve it via GET /repos/{SOURCE_REPO}/milestones and match by title)\n- assignee={assignee} if --assignee was provided (if @me, first resolve your username via `GET /user`)\n\nIMPORTANT: The GitHub Issues API also returns pull requests. Filter them out — exclude any item where pull_request key exists in the response object.\n\nIf in watch mode: Also filter out any issue numbers already in the PROCESSED_ISSUES set from previous batches.\n\nError handling:\n\n- If curl returns an HTTP 401 or 403 → stop and tell the user:\n > \"GitHub authentication failed. Please check your apiKey in the OpenClaw dashboard or in the active OpenClaw config path (`$OPENCLAW_CONFIG_PATH`, default `~/.openclaw/openclaw.json`) under `skills.entries.gh-issues`.\"\n- If the response is an empty array (after filtering) → report \"No issues found matching filters\" and stop (or loop back if in watch mode).\n- If curl fails or returns any other error → report the error verbatim and stop.\n\nParse the JSON response. For each issue, extract: number, title, body, labels (array of label names), assignees, html_url.\n\n---\n\n## Phase 3 — Present & Confirm\n\nDisplay a markdown table of fetched issues:\n\n| # | Title | Labels |\n| --- | ----------------------------- | ------------- |\n| 42 | Fix null pointer in parser | bug, critical |\n| 37 | Add retry logic for API calls | enhancement |\n\nIf FORK_MODE is active, also display:\n\n> \"Fork mode: branches will be pushed to {PUSH_REPO}, PRs will target `{SOURCE_REPO}`\"\n\nIf `--dry-run` is active:\n\n- Display the table and stop. Do not proceed to Phase 4.\n\nIf `--yes` is active:\n\n- Display the table for visibility\n- Auto-process ALL listed issues without asking for confirmation\n- Proceed directly to Phase 4\n\nOtherwise:\nAsk the user to confirm which issues to process:\n\n- \"all\" — process every listed issue\n- Comma-separated numbers (e.g. `42, 37`) — process only those\n- \"cancel\" — abort entirely\n\nWait for user response before proceeding.\n\nWatch mode note: On the first poll, always confirm with the user (unless --yes is set). On subsequent polls, auto-process all new issues without re-confirming (the user already opted in). Still display the table so they can see what's being processed.\n\n---\n\n## Phase 4 — Pre-flight Checks\n\nRun these checks sequentially via exec:\n\n1. **Dirty working tree check:**\n\n ```\n git status --porcelain\n ```\n\n If output is non-empty, warn the user:\n\n > \"Working tree has uncommitted changes. Sub-agents will create branches from HEAD — uncommitted changes will NOT be included. Continue?\"\n > Wait for confirmation. If declined, stop.\n\n2. **Record base branch:**\n\n ```\n git rev-parse --abbrev-ref HEAD\n ```\n\n Store as BASE_BRANCH.\n\n3. **Verify remote access:**\n If FORK_MODE:\n - Verify the fork remote exists. Check if a git remote named `fork` exists:\n ```\n git remote get-url fork\n ```\n If it doesn't exist, add it:\n ```\n git remote add fork https://x-access-token:$GH_TOKEN@github.com/{PUSH_REPO}.git\n ```\n - Also verify origin (the source repo) is reachable:\n ```\n git ls-remote --exit-code origin HEAD\n ```\n\n If not FORK_MODE:\n\n ```\n git ls-remote --exit-code origin HEAD\n ```\n\n If this fails, stop with: \"Cannot reach remote origin. Check your network and git config.\"\n\n4. **Verify GH_TOKEN validity:**\n\n ```\n curl -s -o /dev/null -w \"%{http_code}\" -H \"Authorization: Bearer $GH_TOKEN\" https://api.github.com/user\n ```\n\n If HTTP status is not 200, stop with:\n\n > \"GitHub authentication failed. Please check your apiKey in the OpenClaw dashboard or in the active OpenClaw config path (`$OPENCLAW_CONFIG_PATH`, default `~/.openclaw/openclaw.json`) under `skills.entries.gh-issues`.\"\n\n5. **Check for existing PRs:**\n For each confirmed issue number N, run:\n\n ```\n curl -s -H \"Authorization: Bearer $GH_TOKEN\" -H \"Accept: application/vnd.github+json\" \\\n \"https://api.github.com/repos/{SOURCE_REPO}/pulls?head={PUSH_REPO_OWNER}:fix/issue-{N}&state=open&per_page=1\"\n ```\n\n (Where PUSH_REPO_OWNER is the owner portion of `PUSH_REPO`)\n If the response array is non-empty, remove that issue from the processing list and report:\n\n > \"Skipping #{N} — PR already exists: {html_url}\"\n\n If all issues are skipped, report and stop (or loop back if in watch mode).\n\n6. **Check for in-progress branches (no PR yet = sub-agent still working):**\n For each remaining issue number N (not already skipped by the PR check above), check if a `fix/issue-{N}` branch exists on the **push repo** (which may be a fork, not origin):\n\n ```\n curl -s -o /dev/null -w \"%{http_code}\" \\\n -H \"Authorization: Bearer $GH_TOKEN\" \\\n \"https://api.github.com/repos/{PUSH_REPO}/branches/fix/issue-{N}\"\n ```\n\n If HTTP 200 → the branch exists on the push repo but no open PR was found for it in step 5. Skip that issue:\n\n > \"Skipping #{N} — branch fix/issue-{N} exists on {PUSH_REPO}, fix likely in progress\"\n\n This check uses the GitHub API instead of `git ls-remote` so it works correctly in fork mode (where branches are pushed to the fork, not origin).\n\n If all issues are skipped after this check, report and stop (or loop back if in watch mode).\n\n7. **Check claim-based in-progress tracking:**\n This prevents duplicate processing when a sub-agent from a previous cron run is still working but hasn't pushed a branch or opened a PR yet.\n\n Read the claims file (create empty `{}` if missing):\n\n ```\n CLAIMS_FILE=\"/data/.clawdbot/gh-issues-claims.json\"\n if [ ! -f \"$CLAIMS_FILE\" ]; then\n mkdir -p /data/.clawdbot\n echo '{}' > \"$CLAIMS_FILE\"\n fi\n ```\n\n Parse the claims file. For each entry, check if the claim timestamp is older than 2 hours. If so, remove it (expired — the sub-agent likely finished or failed silently). Write back the cleaned file:\n\n ```\n CLAIMS=$(cat \"$CLAIMS_FILE\")\n CUTOFF=$(date -u -d '2 hours ago' +%Y-%m-%dT%H:%M:%SZ 2>/dev/null || date -u -v-2H +%Y-%m-%dT%H:%M:%SZ)\n CLAIMS=$(echo \"$CLAIMS\" | jq --arg cutoff \"$CUTOFF\" 'to_entries | map(select(.value > $cutoff)) | from_entries')\n echo \"$CLAIMS\" > \"$CLAIMS_FILE\"\n ```\n\n For each remaining issue number N (not already skipped by steps 5 or 6), check if `{SOURCE_REPO}#{N}` exists as a key in the claims file.\n\n If claimed and not expired → skip:\n\n > \"Skipping #{N} — sub-agent claimed this issue {minutes}m ago, still within timeout window\"\n\n Where `{minutes}` is calculated from the claim timestamp to now.\n\n If all issues are skipped after this check, report and stop (or loop back if in watch mode).\n\n---\n\n## Phase 5 — Spawn Sub-agents (Parallel)\n\n**Cron mode (`--cron` is active):**\n\n- **Sequential cursor tracking:** Use a cursor file to track which issue to process next:\n\n ```\n CURSOR_FILE=\"/data/.clawdbot/gh-issues-cursor-{SOURCE_REPO_SLUG}.json\"\n # SOURCE_REPO_SLUG = owner-repo with slashes replaced by hyphens (e.g., openclaw-openclaw)\n ```\n\n Read the cursor file (create if missing):\n\n ```\n if [ ! -f \"$CURSOR_FILE\" ]; then\n echo '{\"last_processed\": null, \"in_progress\": null}' > \"$CURSOR_FILE\"\n fi\n ```\n\n - `last_processed`: issue number of the last completed issue (or null if none)\n - `in_progress`: issue number currently being processed (or null)\n\n- **Select next issue:** Filter the fetched issues list to find the first issue where:\n - Issue number > last_processed (if last_processed is set)\n - AND issue is not in the claims file (not already in progress)\n - AND no PR exists for the issue (checked in Phase 4 step 5)\n - AND no branch exists on the push repo (checked in Phase 4 step 6)\n- If no eligible issue is found after the last_processed cursor, wrap around to the beginning (start from the oldest eligible issue).\n\n- If an eligible issue is found:\n 1. Mark it as in_progress in the cursor file\n 2. Spawn a single sub-agent for that one issue with `cleanup: \"keep\"` and `runTimeoutSeconds: 3600`\n 3. If `--model` was provided, include `model: \"{MODEL}\"` in the spawn config\n 4. If `--notify-channel` was provided, include the channel in the task so the sub-agent can notify\n 5. Do NOT await the sub-agent result — fire and forget\n 6. **Write claim:** After spawning, read the claims file, add `{SOURCE_REPO}#{N}` with the current ISO timestamp, and write it back\n 7. Immediately report: \"Spawned fix agent for #{N} — will create PR when complete\"\n 8. Exit the skill. Do not proceed to Results Collection or Phase 6.\n\n- If no eligible issue is found (all issues either have PRs, have branches, or are in progress), report \"No eligible issues to process — all issues have PRs/branches or are in progress\" and exit.\n\n**Normal mode (`--cron` is NOT active):**\nFor each confirmed issue, spawn a sub-agent using sessions_spawn. Launch up to 8 concurrently (matching `subagents.maxConcurrent: 8`). If more than 8 issues, batch them — launch the next agent as each completes.\n\n**Write claims:** After spawning each sub-agent, read the claims file, add `{SOURCE_REPO}#{N}` with the current ISO timestamp, and write it back (same procedure as cron mode above). This covers interactive usage where watch mode might overlap with cron runs.\n\n### Sub-agent Task Prompt\n\nFor each issue, construct the following prompt and pass it to sessions_spawn. Variables to inject into the template:\n\n- {SOURCE_REPO} — upstream repo where the issue lives\n- {PUSH_REPO} — repo to push branches to (same as SOURCE_REPO unless fork mode)\n- {FORK_MODE} — true/false\n- {PUSH_REMOTE} — `fork` if FORK_MODE, otherwise `origin`\n- {number}, {title}, {url}, {labels}, {body} — from the issue\n- {BASE_BRANCH} — from Phase 4\n- {notify_channel} — Telegram channel ID for notifications (empty if not set). Replace {notify_channel} in the template below with the value of `--notify-channel` flag (or leave as empty string if not provided).\n\nWhen constructing the task, replace all template variables including {notify_channel} with actual values.\n\n```\nYou are a focused code-fix agent. Your task is to fix a single GitHub issue and open a PR.\n\nIMPORTANT: Do NOT use the gh CLI — it is not installed. Use curl with the GitHub REST API for all GitHub operations.\n\nFirst, ensure GH_TOKEN is set. Check: `echo $GH_TOKEN`. If empty, read from config:\nCONFIG_PATH=\"${OPENCLAW_CONFIG_PATH:-${OPENCLAW_STATE_DIR:-$HOME/.openclaw}/openclaw.json}\"\nGH_TOKEN=$(cat \"$CONFIG_PATH\" 2>/dev/null | jq -r '.skills.entries[\"gh-issues\"].apiKey // empty') || GH_TOKEN=$(cat /data/.clawdbot/openclaw.json 2>/dev/null | jq -r '.skills.entries[\"gh-issues\"].apiKey // empty')\n\nUse the token in all GitHub API calls:\ncurl -s -H \"Authorization: Bearer $GH_TOKEN\" -H \"Accept: application/vnd.github+json\" ...\n\n\nSource repo (issues): {SOURCE_REPO}\nPush repo (branches + PRs): {PUSH_REPO}\nFork mode: {FORK_MODE}\nPush remote name: {PUSH_REMOTE}\nBase branch: {BASE_BRANCH}\nNotify channel: {notify_channel}\n\n\n\nRepository: {SOURCE_REPO}\nIssue: #{number}\nTitle: {title}\nURL: {url}\nLabels: {labels}\nBody: {body}\n\n\n\nFollow these steps in order. If any step fails, report the failure and stop.\n\n0. SETUP — Ensure GH_TOKEN is available:\n```\n\nexport GH_TOKEN=$(node -e \"const fs=require('fs'); const c=JSON.parse(fs.readFileSync('/data/.clawdbot/openclaw.json','utf8')); console.log(c.skills?.entries?.['gh-issues']?.apiKey || '')\")\n\n```\nIf that fails, also try:\n```\n\nexport CONFIG_PATH=\"${OPENCLAW_CONFIG_PATH:-${OPENCLAW_STATE_DIR:-$HOME/.openclaw}/openclaw.json}\"\nexport GH_TOKEN=$(cat \"$CONFIG_PATH\" 2>/dev/null | node -e \"const fs=require('fs');const d=JSON.parse(fs.readFileSync(0,'utf8'));console.log(d.skills?.entries?.['gh-issues']?.apiKey||'')\")\n\n```\nVerify: echo \"Token: ${GH_TOKEN:0:10}...\"\n\n1. CONFIDENCE CHECK — Before implementing, assess whether this issue is actionable:\n- Read the issue body carefully. Is the problem clearly described?\n- Search the codebase (grep/find) for the relevant code. Can you locate it?\n- Is the scope reasonable? (single file/function = good, whole subsystem = bad)\n- Is a specific fix suggested or is it a vague complaint?\n\nRate your confidence (1-10). If confidence < 7, STOP and report:\n> \"Skipping #{number}: Low confidence (score: N/10) — [reason: vague requirements | cannot locate code | scope too large | no clear fix suggested]\"\n\nOnly proceed if confidence >= 7.\n\n1. UNDERSTAND — Read the issue carefully. Identify what needs to change and where.\n\n2. BRANCH — Create a feature branch from the base branch:\ngit checkout -b fix/issue-{number} {BASE_BRANCH}\n\n3. ANALYZE — Search the codebase to find relevant files:\n- Use grep/find via exec to locate code related to the issue\n- Read the relevant files to understand the current behavior\n- Identify the root cause\n\n4. IMPLEMENT — Make the minimal, focused fix:\n- Follow existing code style and conventions\n- Change only what is necessary to fix the issue\n- Do not add unrelated changes or new dependencies without justification\n\n5. TEST — Discover and run the existing test suite if one exists:\n- Look for package.json scripts, Makefile targets, pytest, cargo test, etc.\n- Run the relevant tests\n- If tests fail after your fix, attempt ONE retry with a corrected approach\n- If tests still fail, report the failure\n\n6. COMMIT — Stage and commit your changes:\ngit add {changed_files}\ngit commit -m \"fix: {short_description}\n\nFixes {SOURCE_REPO}#{number}\"\n\n7. PUSH — Push the branch:\nFirst, ensure the push remote uses token auth and disable credential helpers:\ngit config --global credential.helper \"\"\ngit remote set-url {PUSH_REMOTE} https://x-access-token:$GH_TOKEN@github.com/{PUSH_REPO}.git\nThen push:\nGIT_ASKPASS=true git push -u {PUSH_REMOTE} fix/issue-{number}\n\n8. PR — Create a pull request using the GitHub API:\n\nIf FORK_MODE is true, the PR goes from your fork to the source repo:\n- head = \"{PUSH_REPO_OWNER}:fix/issue-{number}\"\n- base = \"{BASE_BRANCH}\"\n- PR is created on {SOURCE_REPO}\n\nIf FORK_MODE is false:\n- head = \"fix/issue-{number}\"\n- base = \"{BASE_BRANCH}\"\n- PR is created on {SOURCE_REPO}\n\ncurl -s -X POST \\\n -H \"Authorization: Bearer $GH_TOKEN\" \\\n -H \"Accept: application/vnd.github+json\" \\\n https://api.github.com/repos/{SOURCE_REPO}/pulls \\\n -d '{\n \"title\": \"fix: {title}\",\n \"head\": \"{head_value}\",\n \"base\": \"{BASE_BRANCH}\",\n \"body\": \"## Summary\\n\\n{one_paragraph_description_of_fix}\\n\\n## Changes\\n\\n{bullet_list_of_changes}\\n\\n## Testing\\n\\n{what_was_tested_and_results}\\n\\nFixes {SOURCE_REPO}#{number}\"\n }'\n\nExtract the `html_url` from the response — this is the PR link.\n\n9. REPORT — Send back a summary:\n- PR URL (the html_url from step 8)\n- Files changed (list)\n- Fix summary (1-2 sentences)\n- Any caveats or concerns\n\n10. NOTIFY (if notify_channel is set) — If {notify_channel} is not empty, send a notification to the Telegram channel:\n```\n\nUse the message tool with:\n\n- action: \"send\"\n- channel: \"telegram\"\n- target: \"{notify_channel}\"\n- message: \"✅ PR Created: {SOURCE_REPO}#{number}\n\n{title}\n\n{pr_url}\n\nFiles changed: {files_changed_list}\"\n\n```\n\n\n\n- No force-push, no modifying the base branch\n- No unrelated changes or gratuitous refactoring\n- No new dependencies without strong justification\n- If the issue is unclear or too complex to fix confidently, report your analysis instead of guessing\n- Do NOT use the gh CLI — it is not available. Use curl + GitHub REST API for all GitHub operations.\n- GH_TOKEN is already in the environment — do NOT prompt for auth\n- Time limit: you have 60 minutes max. Be thorough — analyze properly, test your fix, don't rush.\n\n```\n\n### Spawn configuration per sub-agent:\n\n- runTimeoutSeconds: 3600 (60 minutes)\n- cleanup: \"keep\" (preserve transcripts for review)\n- If `--model` was provided, include `model: \"{MODEL}\"` in the spawn config\n\n### Timeout Handling\n\nIf a sub-agent exceeds 60 minutes, record it as:\n\n> \"#{N} — Timed out (issue may be too complex for auto-fix)\"\n\n---\n\n## Results Collection\n\n**If `--cron` is active:** Skip this section entirely — the orchestrator already exited after spawning in Phase 5.\n\nAfter ALL sub-agents complete (or timeout), collect their results. Store the list of successfully opened PRs in `OPEN_PRS` (PR number, branch name, issue number, PR URL) for use in Phase 6.\n\nPresent a summary table:\n\n| Issue | Status | PR | Notes |\n| --------------------- | --------- | ------------------------------ | ------------------------------ |\n| #42 Fix null pointer | PR opened | https://github.com/.../pull/99 | 3 files changed |\n| #37 Add retry logic | Failed | -- | Could not identify target code |\n| #15 Update docs | Timed out | -- | Too complex for auto-fix |\n| #8 Fix race condition | Skipped | -- | PR already exists |\n\n**Status values:**\n\n- **PR opened** — success, link to PR\n- **Failed** — sub-agent could not complete (include reason in Notes)\n- **Timed out** — exceeded 60-minute limit\n- **Skipped** — existing PR detected in pre-flight\n\nEnd with a one-line summary:\n\n> \"Processed {N} issues: {success} PRs opened, {failed} failed, {skipped} skipped.\"\n\n**Send notification to channel (if --notify-channel is set):**\nIf `--notify-channel` was provided, send the final summary to that Telegram channel using the `message` tool:\n\n```\nUse the message tool with:\n- action: \"send\"\n- channel: \"telegram\"\n- target: \"{notify-channel}\"\n- message: \"✅ GitHub Issues Processed\n\nProcessed {N} issues: {success} PRs opened, {failed} failed, {skipped} skipped.\n\n{PR_LIST}\"\n\nWhere PR_LIST includes only successfully opened PRs in format:\n• #{issue_number}: {PR_url} ({notes})\n```\n\nThen proceed to Phase 6.\n\n---\n\n## Phase 6 — PR Review Handler\n\nThis phase monitors open PRs (created by this skill or pre-existing `fix/issue-*` PRs) for review comments and spawns sub-agents to address them.\n\n**When this phase runs:**\n\n- After Results Collection (Phases 2-5 completed) — checks PRs that were just opened\n- When `--reviews-only` flag is set — skips Phases 2-5 entirely, runs only this phase\n- In watch mode — runs every poll cycle after checking for new issues\n\n**Cron review mode (`--cron --reviews-only`):**\nWhen both `--cron` and `--reviews-only` are set:\n\n1. Run token resolution (Phase 2 token section)\n2. Discover open `fix/issue-*` PRs (Step 6.1)\n3. Fetch review comments (Step 6.2)\n4. **Analyze comment content for actionability** (Step 6.3)\n5. If actionable comments are found, spawn ONE review-fix sub-agent for the first PR with unaddressed comments — fire-and-forget (do NOT await result)\n - Use `cleanup: \"keep\"` and `runTimeoutSeconds: 3600`\n - If `--model` was provided, include `model: \"{MODEL}\"` in the spawn config\n6. Report: \"Spawned review handler for PR #{N} — will push fixes when complete\"\n7. Exit the skill immediately. Do not proceed to Step 6.5 (Review Results).\n\nIf no actionable comments found, report \"No actionable review comments found\" and exit.\n\n**Normal mode (non-cron) continues below:**\n\n### Step 6.1 — Discover PRs to Monitor\n\nCollect PRs to check for review comments:\n\n**If coming from Phase 5:** Use the `OPEN_PRS` list from Results Collection.\n\n**If `--reviews-only` or subsequent watch cycle:** Fetch all open PRs with `fix/issue-` branch pattern:\n\n```\ncurl -s -H \"Authorization: Bearer $GH_TOKEN\" -H \"Accept: application/vnd.github+json\" \\\n \"https://api.github.com/repos/{SOURCE_REPO}/pulls?state=open&per_page=100\"\n```\n\nFilter to only PRs where `head.ref` starts with `fix/issue-`.\n\nFor each PR, extract: `number` (PR number), `head.ref` (branch name), `html_url`, `title`, `body`.\n\nIf no PRs found, report \"No open fix/ PRs to monitor\" and stop (or loop back if in watch mode).\n\n### Step 6.2 — Fetch All Review Sources\n\nFor each PR, fetch reviews from multiple sources:\n\n**Fetch PR reviews:**\n\n```\ncurl -s -H \"Authorization: Bearer $GH_TOKEN\" -H \"Accept: application/vnd.github+json\" \\\n \"https://api.github.com/repos/{SOURCE_REPO}/pulls/{pr_number}/reviews\"\n```\n\n**Fetch PR review comments (inline/file-level):**\n\n```\ncurl -s -H \"Authorization: Bearer $GH_TOKEN\" -H \"Accept: application/vnd.github+json\" \\\n \"https://api.github.com/repos/{SOURCE_REPO}/pulls/{pr_number}/comments\"\n```\n\n**Fetch PR issue comments (general conversation):**\n\n```\ncurl -s -H \"Authorization: Bearer $GH_TOKEN\" -H \"Accept: application/vnd.github+json\" \\\n \"https://api.github.com/repos/{SOURCE_REPO}/issues/{pr_number}/comments\"\n```\n\n**Fetch PR body for embedded reviews:**\nSome review tools (like Greptile) embed their feedback directly in the PR body. Check for:\n\n- `` markers\n- Other structured review sections in the PR body\n\n```\ncurl -s -H \"Authorization: Bearer $GH_TOKEN\" -H \"Accept: application/vnd.github+json\" \\\n \"https://api.github.com/repos/{SOURCE_REPO}/pulls/{pr_number}\"\n```\n\nExtract the `body` field and parse for embedded review content.\n\n### Step 6.3 — Analyze Comments for Actionability\n\n**Determine the bot's own username** for filtering:\n\n```\ncurl -s -H \"Authorization: Bearer $GH_TOKEN\" https://api.github.com/user | jq -r '.login'\n```\n\nStore as `BOT_USERNAME`. Exclude any comment where `user.login` equals `BOT_USERNAME`.\n\n**For each comment/review, analyze the content to determine if it requires action:**\n\n**NOT actionable (skip):**\n\n- Pure approvals or \"LGTM\" without suggestions\n- Bot comments that are informational only (CI status, auto-generated summaries without specific requests)\n- Comments already addressed (check if bot replied with \"Addressed in commit...\")\n- Reviews with state `APPROVED` and no inline comments requesting changes\n\n**IS actionable (requires attention):**\n\n- Reviews with state `CHANGES_REQUESTED`\n- Reviews with state `COMMENTED` that contain specific requests:\n - \"this test needs to be updated\"\n - \"please fix\", \"change this\", \"update\", \"can you\", \"should be\", \"needs to\"\n - \"will fail\", \"will break\", \"causes an error\"\n - Mentions of specific code issues (bugs, missing error handling, edge cases)\n- Inline review comments pointing out issues in the code\n- Embedded reviews in PR body that identify:\n - Critical issues or breaking changes\n - Test failures expected\n - Specific code that needs attention\n - Confidence scores with concerns\n\n**Parse embedded review content (e.g., Greptile):**\nLook for sections marked with `` or similar. Extract:\n\n- Summary text\n- Any mentions of \"Critical issue\", \"needs attention\", \"will fail\", \"test needs to be updated\"\n- Confidence scores below 4/5 (indicates concerns)\n\n**Build actionable_comments list** with:\n\n- Source (review, inline comment, PR body, etc.)\n- Author\n- Body text\n- For inline: file path and line number\n- Specific action items identified\n\nIf no actionable comments found across any PR, report \"No actionable review comments found\" and stop (or loop back if in watch mode).\n\n### Step 6.4 — Present Review Comments\n\nDisplay a table of PRs with pending actionable comments:\n\n```\n| PR | Branch | Actionable Comments | Sources |\n|----|--------|---------------------|---------|\n| #99 | fix/issue-42 | 2 comments | @reviewer1, greptile |\n| #101 | fix/issue-37 | 1 comment | @reviewer2 |\n```\n\nIf `--yes` is NOT set and this is not a subsequent watch poll: ask the user to confirm which PRs to address (\"all\", comma-separated PR numbers, or \"skip\").\n\n### Step 6.5 — Spawn Review Fix Sub-agents (Parallel)\n\nFor each PR with actionable comments, spawn a sub-agent. Launch up to 8 concurrently.\n\n**Review fix sub-agent prompt:**\n\n```\nYou are a PR review handler agent. Your task is to address review comments on a pull request by making the requested changes, pushing updates, and replying to each comment.\n\nIMPORTANT: Do NOT use the gh CLI — it is not installed. Use curl with the GitHub REST API for all GitHub operations.\n\nFirst, ensure GH_TOKEN is set. Check: echo $GH_TOKEN. If empty, read from config:\nCONFIG_PATH=\"${OPENCLAW_CONFIG_PATH:-${OPENCLAW_STATE_DIR:-$HOME/.openclaw}/openclaw.json}\"\nGH_TOKEN=$(cat \"$CONFIG_PATH\" 2>/dev/null | jq -r '.skills.entries[\"gh-issues\"].apiKey // empty') || GH_TOKEN=$(cat /data/.clawdbot/openclaw.json 2>/dev/null | jq -r '.skills.entries[\"gh-issues\"].apiKey // empty')\n\n\nRepository: {SOURCE_REPO}\nPush repo: {PUSH_REPO}\nFork mode: {FORK_MODE}\nPush remote: {PUSH_REMOTE}\nPR number: {pr_number}\nPR URL: {pr_url}\nBranch: {branch_name}\n\n\n\n{json_array_of_actionable_comments}\n\nEach comment has:\n- id: comment ID (for replying)\n- user: who left it\n- body: the comment text\n- path: file path (for inline comments)\n- line: line number (for inline comments)\n- diff_hunk: surrounding diff context (for inline comments)\n- source: where the comment came from (review, inline, pr_body, greptile, etc.)\n\n\n\nFollow these steps in order:\n\n0. SETUP — Ensure GH_TOKEN is available:\n```\n\nexport GH_TOKEN=$(node -e \"const fs=require('fs'); const c=JSON.parse(fs.readFileSync('/data/.clawdbot/openclaw.json','utf8')); console.log(c.skills?.entries?.['gh-issues']?.apiKey || '')\")\n\n```\nVerify: echo \"Token: ${GH_TOKEN:0:10}...\"\n\n1. CHECKOUT — Switch to the PR branch:\ngit fetch {PUSH_REMOTE} {branch_name}\ngit checkout {branch_name}\ngit pull {PUSH_REMOTE} {branch_name}\n\n2. UNDERSTAND — Read ALL review comments carefully. Group them by file. Understand what each reviewer is asking for.\n\n3. IMPLEMENT — For each comment, make the requested change:\n- Read the file and locate the relevant code\n- Make the change the reviewer requested\n- If the comment is vague or you disagree, still attempt a reasonable fix but note your concern\n- If the comment asks for something impossible or contradictory, skip it and explain why in your reply\n\n4. TEST — Run existing tests to make sure your changes don't break anything:\n- If tests fail, fix the issue or revert the problematic change\n- Note any test failures in your replies\n\n5. COMMIT — Stage and commit all changes in a single commit:\ngit add {changed_files}\ngit commit -m \"fix: address review comments on PR #{pr_number}\n\nAddresses review feedback from {reviewer_names}\"\n\n6. PUSH — Push the updated branch:\ngit config --global credential.helper \"\"\ngit remote set-url {PUSH_REMOTE} https://x-access-token:$GH_TOKEN@github.com/{PUSH_REPO}.git\nGIT_ASKPASS=true git push {PUSH_REMOTE} {branch_name}\n\n7. REPLY — For each addressed comment, post a reply:\n\nFor inline review comments (have a path/line), reply to the comment thread:\ncurl -s -X POST \\\n -H \"Authorization: Bearer $GH_TOKEN\" \\\n -H \"Accept: application/vnd.github+json\" \\\n https://api.github.com/repos/{SOURCE_REPO}/pulls/{pr_number}/comments/{comment_id}/replies \\\n -d '{\"body\": \"Addressed in commit {short_sha} — {brief_description_of_change}\"}'\n\nFor general PR comments (issue comments), reply on the PR:\ncurl -s -X POST \\\n -H \"Authorization: Bearer $GH_TOKEN\" \\\n -H \"Accept: application/vnd.github+json\" \\\n https://api.github.com/repos/{SOURCE_REPO}/issues/{pr_number}/comments \\\n -d '{\"body\": \"Addressed feedback from @{reviewer}:\\n\\n{summary_of_changes_made}\\n\\nUpdated in commit {short_sha}\"}'\n\nFor comments you could NOT address, reply explaining why:\n\"Unable to address this comment: {reason}. This may need manual review.\"\n\n8. REPORT — Send back a summary:\n- PR URL\n- Number of comments addressed vs skipped\n- Commit SHA\n- Files changed\n- Any comments that need manual attention\n\n\n\n- Only modify files relevant to the review comments\n- Do not make unrelated changes\n- Do not force-push — always regular push\n- If a comment contradicts another comment, address the most recent one and flag the conflict\n- Do NOT use the gh CLI — use curl + GitHub REST API\n- GH_TOKEN is already in the environment — do not prompt for auth\n- Time limit: 60 minutes max\n\n```\n\n**Spawn configuration per sub-agent:**\n\n- runTimeoutSeconds: 3600 (60 minutes)\n- cleanup: \"keep\" (preserve transcripts for review)\n- If `--model` was provided, include `model: \"{MODEL}\"` in the spawn config\n\n### Step 6.6 — Review Results\n\nAfter all review sub-agents complete, present a summary:\n\n```\n| PR | Comments Addressed | Comments Skipped | Commit | Status |\n|----|-------------------|-----------------|--------|--------|\n| #99 fix/issue-42 | 3 | 0 | abc123f | All addressed |\n| #101 fix/issue-37 | 1 | 1 | def456a | 1 needs manual review |\n```\n\nAdd comment IDs from this batch to `ADDRESSED_COMMENTS` set to prevent re-processing.\n\n---\n\n## Watch Mode (if --watch is active)\n\nAfter presenting results from the current batch:\n\n1. Add all issue numbers from this batch to the running set PROCESSED_ISSUES.\n2. Add all addressed comment IDs to ADDRESSED_COMMENTS.\n3. Tell the user:\n > \"Next poll in {interval} minutes... (say 'stop' to end watch mode)\"\n4. Sleep for {interval} minutes.\n5. Go back to **Phase 2 — Fetch Issues**. The fetch will automatically filter out:\n - Issues already in PROCESSED_ISSUES\n - Issues that have existing fix/issue-{N} PRs (caught in Phase 4 pre-flight)\n6. After Phases 2-5 (or if no new issues), run **Phase 6** to check for new review comments on ALL tracked PRs (both newly created and previously opened).\n7. If no new issues AND no new actionable review comments → report \"No new activity. Polling again in {interval} minutes...\" and loop back to step 4.\n8. The user can say \"stop\" at any time to exit watch mode. When stopping, present a final cumulative summary of ALL batches — issues processed AND review comments addressed.\n\n**Context hygiene between polls — IMPORTANT:**\nOnly retain between poll cycles:\n\n- PROCESSED_ISSUES (set of issue numbers)\n- ADDRESSED_COMMENTS (set of comment IDs)\n- OPEN_PRS (list of tracked PRs: number, branch, URL)\n- Cumulative results (one line per issue + one line per review batch)\n- Parsed arguments from Phase 1\n- BASE_BRANCH, SOURCE_REPO, PUSH_REPO, FORK_MODE, BOT_USERNAME\n Do NOT retain issue bodies, comment bodies, sub-agent transcripts, or codebase analysis between polls.\n","skills/gifgrep/SKILL.md":"---\nname: gifgrep\ndescription: Search GIF providers with CLI/TUI, download results, and extract stills/sheets.\nhomepage: https://gifgrep.com\nmetadata:\n {\n \"openclaw\":\n {\n \"emoji\": \"🧲\",\n \"requires\": { \"bins\": [\"gifgrep\"] },\n \"install\":\n [\n {\n \"id\": \"brew\",\n \"kind\": \"brew\",\n \"formula\": \"steipete/tap/gifgrep\",\n \"bins\": [\"gifgrep\"],\n \"label\": \"Install gifgrep (brew)\",\n },\n {\n \"id\": \"go\",\n \"kind\": \"go\",\n \"module\": \"github.com/steipete/gifgrep/cmd/gifgrep@latest\",\n \"bins\": [\"gifgrep\"],\n \"label\": \"Install gifgrep (go)\",\n },\n ],\n },\n }\n---\n\n# gifgrep\n\nUse `gifgrep` to search GIF providers (Tenor/Giphy), browse in a TUI, download results, and extract stills or sheets.\n\nGIF-Grab (gifgrep workflow)\n\n- Search → preview → download → extract (still/sheet) for fast review and sharing.\n\nQuick start\n\n- `gifgrep cats --max 5`\n- `gifgrep cats --format url | head -n 5`\n- `gifgrep search --json cats | jq '.[0].url'`\n- `gifgrep tui \"office handshake\"`\n- `gifgrep cats --download --max 1 --format url`\n\nTUI + previews\n\n- TUI: `gifgrep tui \"query\"`\n- CLI still previews: `--thumbs` (Kitty/Ghostty only; still frame)\n\nDownload + reveal\n\n- `--download` saves to `~/Downloads`\n- `--reveal` shows the last download in Finder\n\nStills + sheets\n\n- `gifgrep still ./clip.gif --at 1.5s -o still.png`\n- `gifgrep sheet ./clip.gif --frames 9 --cols 3 -o sheet.png`\n- Sheets = single PNG grid of sampled frames (great for quick review, docs, PRs, chat).\n- Tune: `--frames` (count), `--cols` (grid width), `--padding` (spacing).\n\nProviders\n\n- `--source auto|tenor|giphy`\n- `GIPHY_API_KEY` required for `--source giphy`\n- `TENOR_API_KEY` optional (Tenor demo key used if unset)\n\nOutput\n\n- `--json` prints an array of results (`id`, `title`, `url`, `preview_url`, `tags`, `width`, `height`)\n- `--format` for pipe-friendly fields (e.g., `url`)\n\nEnvironment tweaks\n\n- `GIFGREP_SOFTWARE_ANIM=1` to force software animation\n- `GIFGREP_CELL_ASPECT=0.5` to tweak preview geometry\n","skills/github/SKILL.md":"---\nname: github\ndescription: Use when interacting with GitHub repositories, authentication, remotes, pushes, or repository creation through the GitHub CLI and related git workflows.\nhomepage: https://cli.github.com/\nmetadata:\n {\n \"emoji\": \"🐙\"\n }\n---\n\n# GitHub\n\n## Overview\n\nPrefer `gh` for all GitHub operations. The default goal is to complete the full GitHub flow with the least possible user effort: detect existing authentication first, reuse it if available, and only ask the user for the smallest missing step.\n\nNever ask the user to manually perform a long GitHub setup if the agent can check, reuse, or finish it directly.\n\n## Core Rules\n\n- Prefer `gh` over raw GitHub API calls.\n- Prefer existing authenticated state over asking for new credentials.\n- Prefer non-interactive checks before prompting the user.\n- Prefer the smallest unblock request:\n - paste a one-time code\n - confirm browser login\n - provide a PAT\n - confirm repository visibility\n- Do not ask the user to manually create the repository if `gh` can do it.\n- Do not ask the user to manually add remotes, initialize Git, or push if the agent can do it.\n- Never print, echo, log, summarize, or restate secrets.\n- Never store secrets in files, commits, repo config, docs, examples, or chat summaries.\n- Never switch credential strategy unless the current one actually fails.\n\n### Interactive login driving rules\n\nWhen using `gh auth login` through a shell session, do not assume the browser login step has started after the first prompt or after sending one key.\n\nYou must continue driving the interactive flow until there is explicit evidence of one of these states:\n\n1. GitHub CLI has clearly started browser/device authentication\n2. GitHub CLI is explicitly waiting for the user to complete external authentication\n3. GitHub CLI reports that authentication succeeded\n4. GitHub CLI reports a real error and cannot continue\n\nWhile driving the login flow:\n\n- keep polling the session after each input\n- handle intermediate prompts such as protocol selection, SSH/HTTPS choice, credential helper confirmation, browser/device flow confirmation, or similar setup questions\n- use minimal inputs such as `enter`, arrow keys, `y`, `n`, or short submitted text as needed\n- do not stop after only one successful prompt response\n- do not tell the user that the browser has opened unless the command output explicitly indicates that a browser login or web flow has started\n- do not tell the user that login is complete unless the command output explicitly indicates success\n\nAcceptable evidence that browser or external auth has actually started includes output such as:\n\n- a message that a browser is being opened\n- a URL to visit\n- a device code to enter\n- explicit wording that GitHub authentication is waiting for completion\n\nIf the shell session remains running but only shows that an earlier prompt was answered, continue polling instead of summarizing the state prematurely.\n\nOnly ask the user to act when the output clearly shows an external action is now required.\n\nWhen the user says they completed browser or device authorization, do not start a new `gh auth login` flow immediately.\n\nInstead:\n\n1. continue polling the existing shell session if it is still running\n2. wait for explicit success or failure from that same session\n3. only if that session has exited or is unrecoverable, then check `gh auth status`\n4. only start a brand-new login flow if the previous login session has clearly failed or been closed\n\nIf the shell output contains prompts like:\n\n- \"Press Enter to open ...\"\n- \"Press Enter to continue\"\n- similar one-step local confirmation prompts\n\nthen send `enter` automatically and continue polling the same session.\n\nDo not stop at that prompt.\nDo not ask the user to confirm completion yet.\nDo not start a new login flow while this session is still running.\n\n## Credential Handling Priority\n\n### 1. Reuse `gh` authentication first\n\nCheck whether `gh` exists and is already authenticated.\n\nUse this path when possible:\n\n- `gh auth status`\n- if authenticated, continue directly with:\n - repo detection\n - `gh repo create`\n - remote setup\n - push\n\nIf `gh` is authenticated, do not ask the user for tokens, passwords, SSH keys, or browser login.\n\n### 2. If `gh` exists but is not authenticated, log in with `gh` directly\n\nIf `gh` is installed but not authenticated, use `gh auth login` directly and continue the GitHub flow after login succeeds.\n\nPreferred order:\n\n1. browser/web login via `gh auth login -h github.com -w`\n2. standard interactive `gh auth login -h github.com`\n3. PAT-based login only if the normal `gh` login path is unavailable or fails\n\nRules:\n\n- do not switch away from `gh` login unless it actually fails\n- do not ask the user to manually create repositories or configure git\n- do not ask the user to perform a long GitHub setup outside the login itself\n- continue automatically as soon as login succeeds\n\nWhen login requires user interaction:\n\n- ask for only the smallest possible action\n- examples:\n - “Please complete the GitHub browser login that just opened.”\n - “Please finish the GitHub CLI login prompts.”\n - “Please paste a GitHub PAT with repo access.”\n\nAfter login succeeds, continue directly with:\n\n- repo detection\n- `gh repo create`\n- remote setup\n- push\n\n### 3. If `gh` is unavailable, try to install it first\n\nIf `gh` is missing:\n\n- first try to install `gh` automatically using the platform's normal package manager or a reasonable official installation path\n- prefer non-interactive installation methods when available\n- after installation, re-check whether `gh` exists and continue with the normal `gh`-first flow\n- only fall back away from `gh` if installation actually fails, is blocked by the environment, or would require disproportionate user effort\n\nAcceptable install attempts include examples like:\n\n- Homebrew on macOS\n- `apt`, `dnf`, or similar package managers on Linux\n- `winget` or other normal package-manager paths on Windows\n\nRules:\n\n- do not ask the user to install `gh` manually before you have tried a reasonable automatic install path\n- do not switch to a different credential strategy just because `gh` is missing initially\n- if installation fails, summarize the blocker briefly and then continue to the fallback path below\n\n### 4. If `gh` is still unavailable, fall back to Git\n\nIf `gh` is missing:\n\n- continue with plain `git`\n- prefer existing Git credential helpers or existing SSH authentication\n- inspect the environment before asking for credentials\n\nCheck for likely reusable auth:\n\n- existing `origin` remote\n- existing Git credential helper\n- working SSH auth to GitHub\n- existing authenticated HTTPS remote behavior\n\nOnly if those paths fail should the user be asked for one minimal credential input.\n\n## Minimal-Interruption Prompting\n\nWhen user input is required, ask for only one concrete item at a time.\n\nGood prompts:\n\n- “Please confirm whether the repo should be public or private.”\n- “Please complete the GitHub login in your browser.”\n- “Please paste a GitHub PAT with repo access.”\n- “Please confirm the GitHub owner/org name.”\n\nBad prompts:\n\n- asking the user to manually create a repo\n- asking the user to configure Git from scratch without first checking\n- asking for multiple credentials at once\n- asking the user to inspect local config manually if the agent can inspect it\n\n## Safe Credential Rules\n\n- Treat PATs, device codes, auth tokens, cookies, and SSH private keys as secrets.\n- Never echo secrets back to the user.\n- Never include secrets in command output summaries.\n- Never write secrets into repo files or examples.\n- Never commit `.env`, token files, or copied credentials.\n- If a command might expose a secret in stdout/stderr, avoid reproducing that output verbatim.\n- If the user gives a token, use it only for the current task unless explicitly told otherwise.\n\n## Standard Execution Flow\n\n### A. Detect environment\n\nCheck in this order:\n\n1. Is this already a Git repo?\n2. Is there already a remote?\n3. Is `gh` installed?\n4. Is `gh` authenticated?\n5. If not using `gh`, does Git already have usable auth?\n\n### B. If already connected to GitHub\n\nIf the repo already has a valid GitHub remote and push access works:\n\n- do not recreate the repo\n- do not replace auth\n- just push\n\n### C. If Git repo exists but no GitHub repo exists yet\n\nPreferred:\n\n- create the GitHub repo with `gh repo create`\n- set or update `origin`\n- push the current branch\n\nOnly ask the user for:\n\n- owner/org if ambiguous\n- public vs private if not specified\n- authentication only if required\n\n### D. If not a Git repo yet\n\nDo the setup directly:\n\n- initialize Git\n- add files\n- create initial commit if needed\n- create GitHub repo\n- add remote\n- push\n\nDo not ask the user to perform these steps manually.\n\n## `gh`-First Behavior\n\nWhen `gh` is available and authenticated, prefer commands in this shape:\n\n- inspect auth\n- create repo\n- set remote if needed\n- push branch\n- optionally enable features like description/homepage/topics later\n\nUse plain Git only for local repository operations or when `gh` cannot complete the GitHub-side action.\n\n## Git / API Fallback Behavior\n\nOnly use this fallback when:\n\n- `gh` is not installed and cannot be installed reasonably\n- or `gh` is installed but unusable for the required operation\n- or the environment blocks the normal `gh` flow in a way that cannot be resolved with minimal user input\n\nFallback order:\n\n1. direct GitHub API + local `git`\n2. plain `git` with existing SSH or credential-helper auth\n3. HTTPS + PAT only if no better reusable auth exists\n\nFallback rules:\n\n- still minimize user effort\n- still prefer reusing existing credentials\n- still never ask the user to create the repository manually if the agent can do it through the API\n- still avoid exposing or persisting secrets\n- still continue automatically once the minimal missing credential or confirmation is provided\n\nIf PAT is required:\n\n- ask only for a token with the minimum needed scope, usually repo write access\n- use it only to complete the push flow\n- avoid leaving it in shell history, files, process args, or remote URLs when possible\n\n## Success Criteria\n\nA successful run means:\n\n- the local project is a valid Git repo if needed\n- the GitHub repo exists\n- the correct remote is configured\n- the current branch is pushed\n- credentials were reused when possible\n- user involvement was limited to the smallest truly necessary action\n- no secret was exposed or persisted unsafely\n","skills/gog/SKILL.md":"---\nname: gog\ndescription: Google Workspace CLI for Gmail, Calendar, Drive, Contacts, Sheets, and Docs.\nhomepage: https://gogcli.sh\nmetadata:\n {\n \"openclaw\":\n {\n \"emoji\": \"🎮\",\n \"requires\": { \"bins\": [\"gog\"] },\n \"install\":\n [\n {\n \"id\": \"brew\",\n \"kind\": \"brew\",\n \"formula\": \"steipete/tap/gogcli\",\n \"bins\": [\"gog\"],\n \"label\": \"Install gog (brew)\",\n },\n ],\n },\n }\n---\n\n# gog\n\nUse `gog` for Gmail/Calendar/Drive/Contacts/Sheets/Docs. Requires OAuth setup.\n\nSetup (once)\n\n- `gog auth credentials /path/to/client_secret.json`\n- `gog auth add you@gmail.com --services gmail,calendar,drive,contacts,docs,sheets`\n- `gog auth list`\n\nCommon commands\n\n- Gmail search: `gog gmail search 'newer_than:7d' --max 10`\n- Gmail messages search (per email, ignores threading): `gog gmail messages search \"in:inbox from:ryanair.com\" --max 20 --account you@example.com`\n- Gmail send (plain): `gog gmail send --to a@b.com --subject \"Hi\" --body \"Hello\"`\n- Gmail send (multi-line): `gog gmail send --to a@b.com --subject \"Hi\" --body-file ./message.txt`\n- Gmail send (stdin): `gog gmail send --to a@b.com --subject \"Hi\" --body-file -`\n- Gmail send (HTML): `gog gmail send --to a@b.com --subject \"Hi\" --body-html \"

Hello

\"`\n- Gmail draft: `gog gmail drafts create --to a@b.com --subject \"Hi\" --body-file ./message.txt`\n- Gmail send draft: `gog gmail drafts send `\n- Gmail reply: `gog gmail send --to a@b.com --subject \"Re: Hi\" --body \"Reply\" --reply-to-message-id `\n- Calendar list events: `gog calendar events --from --to `\n- Calendar create event: `gog calendar create --summary \"Title\" --from --to `\n- Calendar create with color: `gog calendar create --summary \"Title\" --from --to --event-color 7`\n- Calendar update event: `gog calendar update --summary \"New Title\" --event-color 4`\n- Calendar show colors: `gog calendar colors`\n- Drive search: `gog drive search \"query\" --max 10`\n- Contacts: `gog contacts list --max 20`\n- Sheets get: `gog sheets get \"Tab!A1:D10\" --json`\n- Sheets update: `gog sheets update \"Tab!A1:B2\" --values-json '[[\"A\",\"B\"],[\"1\",\"2\"]]' --input USER_ENTERED`\n- Sheets append: `gog sheets append \"Tab!A:C\" --values-json '[[\"x\",\"y\",\"z\"]]' --insert INSERT_ROWS`\n- Sheets clear: `gog sheets clear \"Tab!A2:Z\"`\n- Sheets metadata: `gog sheets metadata --json`\n- Docs export: `gog docs export --format txt --out /tmp/doc.txt`\n- Docs cat: `gog docs cat `\n\nCalendar Colors\n\n- Use `gog calendar colors` to see all available event colors (IDs 1-11)\n- Add colors to events with `--event-color ` flag\n- Event color IDs (from `gog calendar colors` output):\n - 1: #a4bdfc\n - 2: #7ae7bf\n - 3: #dbadff\n - 4: #ff887c\n - 5: #fbd75b\n - 6: #ffb878\n - 7: #46d6db\n - 8: #e1e1e1\n - 9: #5484ed\n - 10: #51b749\n - 11: #dc2127\n\nEmail Formatting\n\n- Prefer plain text. Use `--body-file` for multi-paragraph messages (or `--body-file -` for stdin).\n- Same `--body-file` pattern works for drafts and replies.\n- `--body` does not unescape `\\n`. If you need inline newlines, use a heredoc or `$'Line 1\\n\\nLine 2'`.\n- Use `--body-html` only when you need rich formatting.\n- HTML tags: `

` for paragraphs, `
` for line breaks, `` for bold, `` for italic, `` for links, `