{"owner":"WenyuChiou","repo":"awesome-agentic-ai-zh","hasSkills":true,"totalSkillsCount":4,"totalTokensCount":3365,"categories":["claude-rule","mcp-config","marketplace","plugin-manifest"],"hasMcp":true,"mcpConfig":{"mcpServers":{"awesome-agentic-ai-zh":{"command":"npx","args":["-y","@modelcontextprotocol/server-awesome-agentic-ai-zh"]}}},"found":["CLAUDE.md",".cursor/mcp.json",".claude-plugin/marketplace.json",".claude-plugin/plugin.json"],"skills":{"CLAUDE.md":"# Project Memory — awesome-agentic-ai-zh\n\n> Standing instructions for any AI agent (Claude, Codex, Gemini) working on this repo. Read this **first** before touching exercises or model recommendations.\n\n## 📍 Repo positioning — read before adding anything\n\n**This repo's role**: **learning roadmap + curated resources + simple illustrative cases.**\n\n**Benchmark for \"what we are NOT\"**: [`datawhalechina/hello-agents`](https://github.com/datawhalechina/hello-agents) is the canonical chapter-length zh-TW depth tutorial (16 production capabilities, chapter format). **We don't compete with it; we route to it.**\n\n**Implications when contributing**:\n\n| Decision | Rule |\n|---|---|\n| New stage-level exercise folder | OK if it adds **a roadmap node + dual-path SDK demo + 1-line punchline**. 70-150 lines starter is the right size. |\n| Expanding a starter beyond ~150 lines | **Push back**. If it's growing into chapter-length, add a 📚 callout pointing to hello-agents instead. |\n| Adding a 5th `extension` to README | Diminishing return. Keep README tight (under ~200 lines); extra depth goes to the 📚 callout. |\n| New resource (lib / paper / tool / framework) | Add it only when it has a clear teaching role, current documentation, a verified license or official source, and enough real adoption for the relevant section. A new third-party GitHub repository must have at least 1,000 stars at review time; official provider docs, standards, model cards, and irreplaceable canonical sources are exempt. Curation is the primary value. |\n| New chapter-length tutorial inside this repo | **Push back**. If the topic deserves chapter-length, the right move is: write a 1-page summary + simple illustrative case + 📚 callout to a canonical source (hello-agents / Anthropic Cookbook / framework's own docs). |\n| Trilingual mirror priority | Freeze zh-TW first, then ship matching en + zh-Hans mirrors in the same public-content PR. A partial mirror blocks shipping. |\n\n**One-line summary**: **route → depth, not reinvent**. Every exercise folder includes a visible route back to its Stage resources or to suitable deeper material.\n\n**Existing examples of this pattern** (verified 2026-09-13):\n\n- Stage 3 / 4 / 6 / 7 example READMEs include visible learning resources and a Stage return route\n- Main README and both locale mirrors state the roadmap position near the purpose section\n- `tracks/cli/` is outline-only on purpose (CLI exercises are bash/markdown/config, not Python SDK; doesn't fit the dual-path frame — that's correct)\n\n## Canonical Ollama models (verified against the user's `ollama list` and the official Ollama library)\n\n| Model tag | When to use | Notes |\n|---|---|---|\n| **`gemma4:e4b`** | Stage 1 + 2 (plain chat, prompt engineering) | Effective 4B params; the official Ollama tag page showed a 9.6 GB download on 2026-08-30. **The `:e4b` tag matters** — NOT `gemma3n:e4b`, NOT `gemma3:4b`, NOT `gemma4:latest`. |\n| **`gemma4:e2b`** | Smaller Stage 1+2 alternative | The official Ollama tag page showed a 7.2 GB download on 2026-08-30; actual memory needs vary by runtime and hardware, so do not promise it runs on every 4 GB machine. |\n| **`qwen2.5:3b`** | Stages 3–6 (tool use / agent / ReAct) | 1.9 GB, **reliable tool-use support** (OpenAI function-calling format), default for the current function-calling exercises |\n| **`qwen3.5:4b`** | Stage 7 (debate / eval / observability / streaming / deploy mechanics) | 3.4 GB official Ollama tag. These exercises do not depend on function calling; this row does not replace the Stage 3–6 tool-use default. |\n| **`llama3.2:3b`** | `qwen2.5:3b` alternative for tool use | 2.0 GB, similar capability |\n| **`mistral-nemo:12b`** | Higher-quality local fallback | 7.1 GB, closer-to-cloud quality |\n\n**Wrong tags I've used in error before** (now fixed across 13 files via `.ai/.../rename_gemma.py`):\n\n- ❌ `gemma3:4b` — older naming, replaced 2026-05-12\n- ❌ `gemma3n:e4b` — wrong family, replaced 2026-05-12\n- ✅ `gemma4:e4b` — correct (per user's Ollama installation screenshot)\n\nIf unsure, ask the user to run `ollama list` and verify.\n\n## Canonical Anthropic models\n\n| Model | Use case | Pricing (per 1M tokens) |\n|---|---|---|\n| **`claude-fable-5-1`** | Highest widely released Claude tier; 1M context, 128K max output, and stronger long-running agentic work | $10 input / $50 output; $0.25 cache read |\n| **`claude-mythos-5-1`** | Same model as Fable 5.1, with access limited to vetted cybersecurity and life-science users | $10 input / $50 output; $0.25 cache read |\n| **`claude-haiku-5-5`** | Current small-model tier for scoped tasks; 1M context / 128K output. Stage 1 exercises 1 and 3 use 5.5; other exercises retain Haiku 4.5 compatibility defaults pending parameter, thinking-block, and token-budget migration | Prompts ≤100K: $0.10 input / $0.50 output; >100K: $0.50 / $2.50 |\n| **`claude-haiku-4-5`** | Existing exercise compatibility baseline, not the latest or cheapest Haiku | $1 input / $5 output |\n| **`claude-sonnet-5-5`** | Production default for new work; review the [Sonnet 5.5 migration guide](https://platform.claude.com/docs/en/models/sonnet-5-5/migration-guide) before changing existing tool-calling examples | $2 input / $10 output; $0.10 cache read |\n| **`claude-opus-5-5`** | Current Opus-class default for most workloads; use Fable 5.1 when evals still fall short | $4 input / $20 output; $0.20 cache read |\n\n## Framing rules (do not violate)\n\n1. **Claude is the canonical / production reference** in documentation positioning.\n2. **Ollama is the practice default** because of cost — students should not be blocked by API fees during learning.\n3. **Every exercise must ship BOTH paths**:\n   - Path A (Ollama, primary practice runnable): keep the exercise title, result,\n     and first action visible. Use `<details markdown=\"1\" open>` only when Path A\n     is the single immediate action and its rendered body is short; otherwise use\n     a closed `<details markdown=\"1\">` block for code and troubleshooting.\n   - Path B (Anthropic, `<details markdown=\"1\">`, optional cloud-quality comparison)\n4. **Every exercise must mention budget explicitly** — single-run cost + total stage cost.\n5. **Local LLMs must appear in any model recommendation list** — never list cloud-only options.\n\n## Exercise file conventions\n\n- `starter.py` = Ollama / OpenAI-compatible default (Path A)\n- `starter_anthropic.py` = Anthropic SDK version (Path B)\n- `test.py` = mock-based tests for the Ollama starter (OpenAI-compat response shape)\n- `test_anthropic.py` = mock-based tests for the Anthropic starter (content-block shape)\n- `requirements.txt` = both `openai` and `anthropic` pinned\n- `README.md` = trilingual switcher + 怎麼跑（兩條 path）+ budget per path + walkthrough + common pitfalls\n- Each starter ends with `# === 自我驗證 ===` block containing 2+ `assert` statements\n- Each Python file headers Windows-cp950 UTF-8 reconfigure:\n  ```python\n  import sys\n  if hasattr(sys.stdout, \"reconfigure\"):\n      sys.stdout.reconfigure(encoding=\"utf-8\", errors=\"replace\")\n  ```\n\n## Translation rules\n\n- **zh-TW canonical** (`.md` without language suffix). zh-Hans + en mirror.\n- Freeze the Traditional Chinese meaning before translation. A bounded translation agent may produce en + zh-Hans only after its file scope, URLs, numbers, headings, and safety boundaries are fixed.\n- Mechanical conversion is only a first pass. Run the Hans, mirror, anchor, and locale-link gates, then do a human-readable semantic comparison.\n\n## Codex delegation rules\n\n- The primary agent owns scope, architecture, governance, final integration, Git, and user communication.\n- Delegated executors receive bounded file ownership, acceptance commands, a return contract, and a stop condition. They must not revert concurrent work.\n- A separate reviewer reads the final stable staged diff. Any later edit invalidates that review fingerprint.\n- Agent boundaries do not create commit boundaries. Stage explicit paths only, run the required gates, and commit the accepted integrated result.\n\n## Current curriculum contract (verified 2026-09-13)\n\n| Component | Status |\n|---|---|\n| Public curriculum | Stage 0–8, Stage 7.5, A1–A3, five role paths, walkthrough, Capstone, Glossary, and core resource pages have Traditional Chinese, Simplified Chinese, and English routes. |\n| Reader path | Enrolled pages keep goals, bold core terms, required reading, rated projects/resources, exercise outcomes, and completion checks visible; setup, long code, alternatives, and troubleshooting may be collapsed. |\n| Examples | Model-backed teaching folders keep a free/local Ollama path, an optional Anthropic path, budget guidance, and offline behavior tests unless the exercise is deliberately model-free. |\n| Stage 5 | The chapter contains five cumulative exercises plus 5.1–5.8 reference entrances; the tool-calling tutor remains the installable meta-example. |\n| Stage 6 | The reader path, advanced RAG/Memory pages, isolated collections, chunk-overlap guard, persistent memory, and offline behavior tests are present. Live model output quality is not claimed. |\n| Stage 7 | The main order is Eval → Observability → Approval/Recovery → Deploy; Multi-Agent remains optional. Six example folders cover the production mechanics. |\n| Automated checks | On 2026-09-13, 58 `scripts/test_*.py` modules collect 1,145 tests. Counts are a dated observation; CI and `pytest --collect-only` are the current source of truth. |\n| Merge gate | `Required / pr-gate` is the stable required check. A green machine gate does not replace maintainer review. |\n",".cursor/mcp.json":"# Model Context Protocol (MCP) Configuration\n\nPath: `.cursor/mcp.json`\n\n[View Raw Content on GitHub](https://raw.githubusercontent.com/WenyuChiou/awesome-agentic-ai-zh/HEAD/.cursor/mcp.json)",".claude-plugin/marketplace.json":"# Claude Plugin Marketplace Catalog\n\nPath: `.claude-plugin/marketplace.json`\n\n[View Raw Content on GitHub](https://raw.githubusercontent.com/WenyuChiou/awesome-agentic-ai-zh/HEAD/.claude-plugin/marketplace.json)",".claude-plugin/plugin.json":"# claude-plugin Plugin Manifest\n\nPath: `.claude-plugin/plugin.json`\n\n[View Raw Content on GitHub](https://raw.githubusercontent.com/WenyuChiou/awesome-agentic-ai-zh/HEAD/.claude-plugin/plugin.json)"},"files":{"CLAUDE.md":"# Project Memory — awesome-agentic-ai-zh\n\n> Standing instructions for any AI agent (Claude, Codex, Gemini) working on this repo. Read this **first** before touching exercises or model recommendations.\n\n## 📍 Repo positioning — read before adding anything\n\n**This repo's role**: **learning roadmap + curated resources + simple illustrative cases.**\n\n**Benchmark for \"what we are NOT\"**: [`datawhalechina/hello-agents`](https://github.com/datawhalechina/hello-agents) is the canonical chapter-length zh-TW depth tutorial (16 production capabilities, chapter format). **We don't compete with it; we route to it.**\n\n**Implications when contributing**:\n\n| Decision | Rule |\n|---|---|\n| New stage-level exercise folder | OK if it adds **a roadmap node + dual-path SDK demo + 1-line punchline**. 70-150 lines starter is the right size. |\n| Expanding a starter beyond ~150 lines | **Push back**. If it's growing into chapter-length, add a 📚 callout pointing to hello-agents instead. |\n| Adding a 5th `extension` to README | Diminishing return. Keep README tight (under ~200 lines); extra depth goes to the 📚 callout. |\n| New resource (lib / paper / tool / framework) | Add it only when it has a clear teaching role, current documentation, a verified license or official source, and enough real adoption for the relevant section. A new third-party GitHub repository must have at least 1,000 stars at review time; official provider docs, standards, model cards, and irreplaceable canonical sources are exempt. Curation is the primary value. |\n| New chapter-length tutorial inside this repo | **Push back**. If the topic deserves chapter-length, the right move is: write a 1-page summary + simple illustrative case + 📚 callout to a canonical source (hello-agents / Anthropic Cookbook / framework's own docs). |\n| Trilingual mirror priority | Freeze zh-TW first, then ship matching en + zh-Hans mirrors in the same public-content PR. A partial mirror blocks shipping. |\n\n**One-line summary**: **route → depth, not reinvent**. Every exercise folder includes a visible route back to its Stage resources or to suitable deeper material.\n\n**Existing examples of this pattern** (verified 2026-09-13):\n\n- Stage 3 / 4 / 6 / 7 example READMEs include visible learning resources and a Stage return route\n- Main README and both locale mirrors state the roadmap position near the purpose section\n- `tracks/cli/` is outline-only on purpose (CLI exercises are bash/markdown/config, not Python SDK; doesn't fit the dual-path frame — that's correct)\n\n## Canonical Ollama models (verified against the user's `ollama list` and the official Ollama library)\n\n| Model tag | When to use | Notes |\n|---|---|---|\n| **`gemma4:e4b`** | Stage 1 + 2 (plain chat, prompt engineering) | Effective 4B params; the official Ollama tag page showed a 9.6 GB download on 2026-08-30. **The `:e4b` tag matters** — NOT `gemma3n:e4b`, NOT `gemma3:4b`, NOT `gemma4:latest`. |\n| **`gemma4:e2b`** | Smaller Stage 1+2 alternative | The official Ollama tag page showed a 7.2 GB download on 2026-08-30; actual memory needs vary by runtime and hardware, so do not promise it runs on every 4 GB machine. |\n| **`qwen2.5:3b`** | Stages 3–6 (tool use / agent / ReAct) | 1.9 GB, **reliable tool-use support** (OpenAI function-calling format), default for the current function-calling exercises |\n| **`qwen3.5:4b`** | Stage 7 (debate / eval / observability / streaming / deploy mechanics) | 3.4 GB official Ollama tag. These exercises do not depend on function calling; this row does not replace the Stage 3–6 tool-use default. |\n| **`llama3.2:3b`** | `qwen2.5:3b` alternative for tool use | 2.0 GB, similar capability |\n| **`mistral-nemo:12b`** | Higher-quality local fallback | 7.1 GB, closer-to-cloud quality |\n\n**Wrong tags I've used in error before** (now fixed across 13 files via `.ai/.../rename_gemma.py`):\n\n- ❌ `gemma3:4b` — older naming, replaced 2026-05-12\n- ❌ `gemma3n:e4b` — wrong family, replaced 2026-05-12\n- ✅ `gemma4:e4b` — correct (per user's Ollama installation screenshot)\n\nIf unsure, ask the user to run `ollama list` and verify.\n\n## Canonical Anthropic models\n\n| Model | Use case | Pricing (per 1M tokens) |\n|---|---|---|\n| **`claude-fable-5-1`** | Highest widely released Claude tier; 1M context, 128K max output, and stronger long-running agentic work | $10 input / $50 output; $0.25 cache read |\n| **`claude-mythos-5-1`** | Same model as Fable 5.1, with access limited to vetted cybersecurity and life-science users | $10 input / $50 output; $0.25 cache read |\n| **`claude-haiku-5-5`** | Current small-model tier for scoped tasks; 1M context / 128K output. Stage 1 exercises 1 and 3 use 5.5; other exercises retain Haiku 4.5 compatibility defaults pending parameter, thinking-block, and token-budget migration | Prompts ≤100K: $0.10 input / $0.50 output; >100K: $0.50 / $2.50 |\n| **`claude-haiku-4-5`** | Existing exercise compatibility baseline, not the latest or cheapest Haiku | $1 input / $5 output |\n| **`claude-sonnet-5-5`** | Production default for new work; review the [Sonnet 5.5 migration guide](https://platform.claude.com/docs/en/models/sonnet-5-5/migration-guide) before changing existing tool-calling examples | $2 input / $10 output; $0.10 cache read |\n| **`claude-opus-5-5`** | Current Opus-class default for most workloads; use Fable 5.1 when evals still fall short | $4 input / $20 output; $0.20 cache read |\n\n## Framing rules (do not violate)\n\n1. **Claude is the canonical / production reference** in documentation positioning.\n2. **Ollama is the practice default** because of cost — students should not be blocked by API fees during learning.\n3. **Every exercise must ship BOTH paths**:\n   - Path A (Ollama, primary practice runnable): keep the exercise title, result,\n     and first action visible. Use `<details markdown=\"1\" open>` only when Path A\n     is the single immediate action and its rendered body is short; otherwise use\n     a closed `<details markdown=\"1\">` block for code and troubleshooting.\n   - Path B (Anthropic, `<details markdown=\"1\">`, optional cloud-quality comparison)\n4. **Every exercise must mention budget explicitly** — single-run cost + total stage cost.\n5. **Local LLMs must appear in any model recommendation list** — never list cloud-only options.\n\n## Exercise file conventions\n\n- `starter.py` = Ollama / OpenAI-compatible default (Path A)\n- `starter_anthropic.py` = Anthropic SDK version (Path B)\n- `test.py` = mock-based tests for the Ollama starter (OpenAI-compat response shape)\n- `test_anthropic.py` = mock-based tests for the Anthropic starter (content-block shape)\n- `requirements.txt` = both `openai` and `anthropic` pinned\n- `README.md` = trilingual switcher + 怎麼跑（兩條 path）+ budget per path + walkthrough + common pitfalls\n- Each starter ends with `# === 自我驗證 ===` block containing 2+ `assert` statements\n- Each Python file headers Windows-cp950 UTF-8 reconfigure:\n  ```python\n  import sys\n  if hasattr(sys.stdout, \"reconfigure\"):\n      sys.stdout.reconfigure(encoding=\"utf-8\", errors=\"replace\")\n  ```\n\n## Translation rules\n\n- **zh-TW canonical** (`.md` without language suffix). zh-Hans + en mirror.\n- Freeze the Traditional Chinese meaning before translation. A bounded translation agent may produce en + zh-Hans only after its file scope, URLs, numbers, headings, and safety boundaries are fixed.\n- Mechanical conversion is only a first pass. Run the Hans, mirror, anchor, and locale-link gates, then do a human-readable semantic comparison.\n\n## Codex delegation rules\n\n- The primary agent owns scope, architecture, governance, final integration, Git, and user communication.\n- Delegated executors receive bounded file ownership, acceptance commands, a return contract, and a stop condition. They must not revert concurrent work.\n- A separate reviewer reads the final stable staged diff. Any later edit invalidates that review fingerprint.\n- Agent boundaries do not create commit boundaries. Stage explicit paths only, run the required gates, and commit the accepted integrated result.\n\n## Current curriculum contract (verified 2026-09-13)\n\n| Component | Status |\n|---|---|\n| Public curriculum | Stage 0–8, Stage 7.5, A1–A3, five role paths, walkthrough, Capstone, Glossary, and core resource pages have Traditional Chinese, Simplified Chinese, and English routes. |\n| Reader path | Enrolled pages keep goals, bold core terms, required reading, rated projects/resources, exercise outcomes, and completion checks visible; setup, long code, alternatives, and troubleshooting may be collapsed. |\n| Examples | Model-backed teaching folders keep a free/local Ollama path, an optional Anthropic path, budget guidance, and offline behavior tests unless the exercise is deliberately model-free. |\n| Stage 5 | The chapter contains five cumulative exercises plus 5.1–5.8 reference entrances; the tool-calling tutor remains the installable meta-example. |\n| Stage 6 | The reader path, advanced RAG/Memory pages, isolated collections, chunk-overlap guard, persistent memory, and offline behavior tests are present. Live model output quality is not claimed. |\n| Stage 7 | The main order is Eval → Observability → Approval/Recovery → Deploy; Multi-Agent remains optional. Six example folders cover the production mechanics. |\n| Automated checks | On 2026-09-13, 58 `scripts/test_*.py` modules collect 1,145 tests. Counts are a dated observation; CI and `pytest --collect-only` are the current source of truth. |\n| Merge gate | `Required / pr-gate` is the stable required check. A green machine gate does not replace maintainer review. |\n",".cursor/mcp.json":"# Model Context Protocol (MCP) Configuration\n\nPath: `.cursor/mcp.json`\n\n[View Raw Content on GitHub](https://raw.githubusercontent.com/WenyuChiou/awesome-agentic-ai-zh/HEAD/.cursor/mcp.json)",".claude-plugin/marketplace.json":"# Claude Plugin Marketplace Catalog\n\nPath: `.claude-plugin/marketplace.json`\n\n[View Raw Content on GitHub](https://raw.githubusercontent.com/WenyuChiou/awesome-agentic-ai-zh/HEAD/.claude-plugin/marketplace.json)",".claude-plugin/plugin.json":"# claude-plugin Plugin Manifest\n\nPath: `.claude-plugin/plugin.json`\n\n[View Raw Content on GitHub](https://raw.githubusercontent.com/WenyuChiou/awesome-agentic-ai-zh/HEAD/.claude-plugin/plugin.json)"},"items":[{"name":"CLAUDE.md","path":"CLAUDE.md","rawUrl":"https://raw.githubusercontent.com/WenyuChiou/awesome-agentic-ai-zh/HEAD/CLAUDE.md","title":"Claude Agent Guidelines & System Prompt","category":"claude-rule","format":"markdown","content":"# Project Memory — awesome-agentic-ai-zh\n\n> Standing instructions for any AI agent (Claude, Codex, Gemini) working on this repo. Read this **first** before touching exercises or model recommendations.\n\n## 📍 Repo positioning — read before adding anything\n\n**This repo's role**: **learning roadmap + curated resources + simple illustrative cases.**\n\n**Benchmark for \"what we are NOT\"**: [`datawhalechina/hello-agents`](https://github.com/datawhalechina/hello-agents) is the canonical chapter-length zh-TW depth tutorial (16 production capabilities, chapter format). **We don't compete with it; we route to it.**\n\n**Implications when contributing**:\n\n| Decision | Rule |\n|---|---|\n| New stage-level exercise folder | OK if it adds **a roadmap node + dual-path SDK demo + 1-line punchline**. 70-150 lines starter is the right size. |\n| Expanding a starter beyond ~150 lines | **Push back**. If it's growing into chapter-length, add a 📚 callout pointing to hello-agents instead. |\n| Adding a 5th `extension` to README | Diminishing return. Keep README tight (under ~200 lines); extra depth goes to the 📚 callout. |\n| New resource (lib / paper / tool / framework) | Add it only when it has a clear teaching role, current documentation, a verified license or official source, and enough real adoption for the relevant section. A new third-party GitHub repository must have at least 1,000 stars at review time; official provider docs, standards, model cards, and irreplaceable canonical sources are exempt. Curation is the primary value. |\n| New chapter-length tutorial inside this repo | **Push back**. If the topic deserves chapter-length, the right move is: write a 1-page summary + simple illustrative case + 📚 callout to a canonical source (hello-agents / Anthropic Cookbook / framework's own docs). |\n| Trilingual mirror priority | Freeze zh-TW first, then ship matching en + zh-Hans mirrors in the same public-content PR. A partial mirror blocks shipping. |\n\n**One-line summary**: **route → depth, not reinvent**. Every exercise folder includes a visible route back to its Stage resources or to suitable deeper material.\n\n**Existing examples of this pattern** (verified 2026-09-13):\n\n- Stage 3 / 4 / 6 / 7 example READMEs include visible learning resources and a Stage return route\n- Main README and both locale mirrors state the roadmap position near the purpose section\n- `tracks/cli/` is outline-only on purpose (CLI exercises are bash/markdown/config, not Python SDK; doesn't fit the dual-path frame — that's correct)\n\n## Canonical Ollama models (verified against the user's `ollama list` and the official Ollama library)\n\n| Model tag | When to use | Notes |\n|---|---|---|\n| **`gemma4:e4b`** | Stage 1 + 2 (plain chat, prompt engineering) | Effective 4B params; the official Ollama tag page showed a 9.6 GB download on 2026-08-30. **The `:e4b` tag matters** — NOT `gemma3n:e4b`, NOT `gemma3:4b`, NOT `gemma4:latest`. |\n| **`gemma4:e2b`** | Smaller Stage 1+2 alternative | The official Ollama tag page showed a 7.2 GB download on 2026-08-30; actual memory needs vary by runtime and hardware, so do not promise it runs on every 4 GB machine. |\n| **`qwen2.5:3b`** | Stages 3–6 (tool use / agent / ReAct) | 1.9 GB, **reliable tool-use support** (OpenAI function-calling format), default for the current function-calling exercises |\n| **`qwen3.5:4b`** | Stage 7 (debate / eval / observability / streaming / deploy mechanics) | 3.4 GB official Ollama tag. These exercises do not depend on function calling; this row does not replace the Stage 3–6 tool-use default. |\n| **`llama3.2:3b`** | `qwen2.5:3b` alternative for tool use | 2.0 GB, similar capability |\n| **`mistral-nemo:12b`** | Higher-quality local fallback | 7.1 GB, closer-to-cloud quality |\n\n**Wrong tags I've used in error before** (now fixed across 13 files via `.ai/.../rename_gemma.py`):\n\n- ❌ `gemma3:4b` — older naming, replaced 2026-05-12\n- ❌ `gemma3n:e4b` — wrong family, replaced 2026-05-12\n- ✅ `gemma4:e4b` — correct (per user's Ollama installation screenshot)\n\nIf unsure, ask the user to run `ollama list` and verify.\n\n## Canonical Anthropic models\n\n| Model | Use case | Pricing (per 1M tokens) |\n|---|---|---|\n| **`claude-fable-5-1`** | Highest widely released Claude tier; 1M context, 128K max output, and stronger long-running agentic work | $10 input / $50 output; $0.25 cache read |\n| **`claude-mythos-5-1`** | Same model as Fable 5.1, with access limited to vetted cybersecurity and life-science users | $10 input / $50 output; $0.25 cache read |\n| **`claude-haiku-5-5`** | Current small-model tier for scoped tasks; 1M context / 128K output. Stage 1 exercises 1 and 3 use 5.5; other exercises retain Haiku 4.5 compatibility defaults pending parameter, thinking-block, and token-budget migration | Prompts ≤100K: $0.10 input / $0.50 output; >100K: $0.50 / $2.50 |\n| **`claude-haiku-4-5`** | Existing exercise compatibility baseline, not the latest or cheapest Haiku | $1 input / $5 output |\n| **`claude-sonnet-5-5`** | Production default for new work; review the [Sonnet 5.5 migration guide](https://platform.claude.com/docs/en/models/sonnet-5-5/migration-guide) before changing existing tool-calling examples | $2 input / $10 output; $0.10 cache read |\n| **`claude-opus-5-5`** | Current Opus-class default for most workloads; use Fable 5.1 when evals still fall short | $4 input / $20 output; $0.20 cache read |\n\n## Framing rules (do not violate)\n\n1. **Claude is the canonical / production reference** in documentation positioning.\n2. **Ollama is the practice default** because of cost — students should not be blocked by API fees during learning.\n3. **Every exercise must ship BOTH paths**:\n   - Path A (Ollama, primary practice runnable): keep the exercise title, result,\n     and first action visible. Use `<details markdown=\"1\" open>` only when Path A\n     is the single immediate action and its rendered body is short; otherwise use\n     a closed `<details markdown=\"1\">` block for code and troubleshooting.\n   - Path B (Anthropic, `<details markdown=\"1\">`, optional cloud-quality comparison)\n4. **Every exercise must mention budget explicitly** — single-run cost + total stage cost.\n5. **Local LLMs must appear in any model recommendation list** — never list cloud-only options.\n\n## Exercise file conventions\n\n- `starter.py` = Ollama / OpenAI-compatible default (Path A)\n- `starter_anthropic.py` = Anthropic SDK version (Path B)\n- `test.py` = mock-based tests for the Ollama starter (OpenAI-compat response shape)\n- `test_anthropic.py` = mock-based tests for the Anthropic starter (content-block shape)\n- `requirements.txt` = both `openai` and `anthropic` pinned\n- `README.md` = trilingual switcher + 怎麼跑（兩條 path）+ budget per path + walkthrough + common pitfalls\n- Each starter ends with `# === 自我驗證 ===` block containing 2+ `assert` statements\n- Each Python file headers Windows-cp950 UTF-8 reconfigure:\n  ```python\n  import sys\n  if hasattr(sys.stdout, \"reconfigure\"):\n      sys.stdout.reconfigure(encoding=\"utf-8\", errors=\"replace\")\n  ```\n\n## Translation rules\n\n- **zh-TW canonical** (`.md` without language suffix). zh-Hans + en mirror.\n- Freeze the Traditional Chinese meaning before translation. A bounded translation agent may produce en + zh-Hans only after its file scope, URLs, numbers, headings, and safety boundaries are fixed.\n- Mechanical conversion is only a first pass. Run the Hans, mirror, anchor, and locale-link gates, then do a human-readable semantic comparison.\n\n## Codex delegation rules\n\n- The primary agent owns scope, architecture, governance, final integration, Git, and user communication.\n- Delegated executors receive bounded file ownership, acceptance commands, a return contract, and a stop condition. They must not revert concurrent work.\n- A separate reviewer reads the final stable staged diff. Any later edit invalidates that review fingerprint.\n- Agent boundaries do not create commit boundaries. Stage explicit paths only, run the required gates, and commit the accepted integrated result.\n\n## Current curriculum contract (verified 2026-09-13)\n\n| Component | Status |\n|---|---|\n| Public curriculum | Stage 0–8, Stage 7.5, A1–A3, five role paths, walkthrough, Capstone, Glossary, and core resource pages have Traditional Chinese, Simplified Chinese, and English routes. |\n| Reader path | Enrolled pages keep goals, bold core terms, required reading, rated projects/resources, exercise outcomes, and completion checks visible; setup, long code, alternatives, and troubleshooting may be collapsed. |\n| Examples | Model-backed teaching folders keep a free/local Ollama path, an optional Anthropic path, budget guidance, and offline behavior tests unless the exercise is deliberately model-free. |\n| Stage 5 | The chapter contains five cumulative exercises plus 5.1–5.8 reference entrances; the tool-calling tutor remains the installable meta-example. |\n| Stage 6 | The reader path, advanced RAG/Memory pages, isolated collections, chunk-overlap guard, persistent memory, and offline behavior tests are present. Live model output quality is not claimed. |\n| Stage 7 | The main order is Eval → Observability → Approval/Recovery → Deploy; Multi-Agent remains optional. Six example folders cover the production mechanics. |\n| Automated checks | On 2026-09-13, 58 `scripts/test_*.py` modules collect 1,145 tests. Counts are a dated observation; CI and `pytest --collect-only` are the current source of truth. |\n| Merge gate | `Required / pr-gate` is the stable required check. A green machine gate does not replace maintainer review. |\n","isInternal":false,"tokens":2465,"sizeBytes":9626},{"name":"marketplace.json","path":".claude-plugin/marketplace.json","rawUrl":"https://raw.githubusercontent.com/WenyuChiou/awesome-agentic-ai-zh/HEAD/.claude-plugin/marketplace.json","title":"Claude Plugin Marketplace Catalog","category":"marketplace","format":"json","content":"# Claude Plugin Marketplace Catalog\n\nPath: `.claude-plugin/marketplace.json`\n\n[View Raw Content on GitHub](https://raw.githubusercontent.com/WenyuChiou/awesome-agentic-ai-zh/HEAD/.claude-plugin/marketplace.json)","tokens":300,"sizeBytes":211},{"name":"plugin.json","path":".claude-plugin/plugin.json","rawUrl":"https://raw.githubusercontent.com/WenyuChiou/awesome-agentic-ai-zh/HEAD/.claude-plugin/plugin.json","title":"claude-plugin Plugin Manifest","category":"plugin-manifest","format":"json","content":"# claude-plugin Plugin Manifest\n\nPath: `.claude-plugin/plugin.json`\n\n[View Raw Content on GitHub](https://raw.githubusercontent.com/WenyuChiou/awesome-agentic-ai-zh/HEAD/.claude-plugin/plugin.json)","tokens":300,"sizeBytes":197},{"name":"mcp.json","path":".cursor/mcp.json","rawUrl":"https://raw.githubusercontent.com/WenyuChiou/awesome-agentic-ai-zh/HEAD/.cursor/mcp.json","title":"Model Context Protocol (MCP) Configuration","category":"mcp-config","format":"json","content":"# Model Context Protocol (MCP) Configuration\n\nPath: `.cursor/mcp.json`\n\n[View Raw Content on GitHub](https://raw.githubusercontent.com/WenyuChiou/awesome-agentic-ai-zh/HEAD/.cursor/mcp.json)","tokens":300,"sizeBytes":190}],"systemPromptSnippet":"<yakaai_skills repo=\"WenyuChiou/awesome-agentic-ai-zh\">\n<!-- File: CLAUDE.md (Tokens: ~2465 | Category: claude-rule) -->\n# Project Memory — awesome-agentic-ai-zh\n\n> Standing instructions for any AI agent (Claude, Codex, Gemini) working on this repo. Read this **first** before touching exercises or model recommendations.\n\n## 📍 Repo positioning — read before adding anything\n\n**This repo's role**: **learning roadmap + curated resources + simple illustrative cases.**\n\n**Benchmark for \"what we are NOT\"**: [`datawhalechina/hello-agents`](https://github.com/datawhalechina/hello-agents) is the canonical chapter-length zh-TW depth tutorial (16 production capabilities, chapter format). **We don't compete with it; we route to it.**\n\n**Implications when contributing**:\n\n| Decision | Rule |\n|---|---|\n| New stage-level exercise folder | OK if it adds **a roadmap node + dual-path SDK demo + 1-line punchline**. 70-150 lines starter is the right size. |\n| Expanding a starter beyond ~150 lines | **Push back**. If it's growing into chapter-length, add a 📚 callout pointing to hello-agents instead. |\n| Adding a 5th `extension` to README | Diminishing return. Keep README tight (under ~200 lines); extra depth goes to the 📚 callout. |\n| New resource (lib / paper / tool / framework) | Add it only when it has a clear teaching role, current documentation, a verified license or official source, and enough real adoption for the relevant section. A new third-party GitHub repository must have at least 1,000 stars at review time; official provider docs, standards, model cards, and irreplaceable canonical sources are exempt. Curation is the primary value. |\n| New chapter-length tutorial inside this repo | **Push back**. If the topic deserves chapter-length, the right move is: write a 1-page summary + simple illustrative case + 📚 callout to a canonical source (hello-agents / Anthropic Cookbook / framework's own docs). |\n| Trilingual mirror priority | Freeze zh-TW first, then ship matching en + zh-Hans mirrors in the same public-content PR. A partial mirror blocks shipping. |\n\n**One-line summary**: **route → depth, not reinvent**. Every exercise folder includes a visible route back to its Stage resources or to suitable deeper material.\n\n**Existing examples of this pattern** (verified 2026-09-13):\n\n- Stage 3 / 4 / 6 / 7 example READMEs include visible learning resources and a Stage return route\n- Main README and both locale mirrors state the roadmap position near the purpose section\n- `tracks/cli/` is outline-only on purpose (CLI exercises are bash/markdown/config, not Python SDK; doesn't fit the dual-path frame — that's correct)\n\n## Canonical Ollama models (verified against the user's `ollama list` and the official Ollama library)\n\n| Model tag | When to use | Notes |\n|---|---|---|\n| **`gemma4:e4b`** | Stage 1 + 2 (plain chat, prompt engineering) | Effective 4B params; the official Ollama tag page showed a 9.6 GB download on 2026-08-30. **The `:e4b` tag matters** — NOT `gemma3n:e4b`, NOT `gemma3:4b`, NOT `gemma4:latest`. |\n| **`gemma4:e2b`** | Smaller Stage 1+2 alternative | The official Ollama tag page showed a 7.2 GB download on 2026-08-30; actual memory needs vary by runtime and hardware, so do not promise it runs on every 4 GB machine. |\n| **`qwen2.5:3b`** | Stages 3–6 (tool use / agent / ReAct) | 1.9 GB, **reliable tool-use support** (OpenAI function-calling format), default for the current function-calling exercises |\n| **`qwen3.5:4b`** | Stage 7 (debate / eval / observability / streaming / deploy mechanics) | 3.4 GB official Ollama tag. These exercises do not depend on function calling; this row does not replace the Stage 3–6 tool-use default. |\n| **`llama3.2:3b`** | `qwen2.5:3b` alternative for tool use | 2.0 GB, similar capability |\n| **`mistral-nemo:12b`** | Higher-quality local fallback | 7.1 GB, closer-to-cloud quality |\n\n**Wrong tags I've used in error before** (now fixed across 13 files via `.ai/.../rename_gemma.py`):\n\n- ❌ `gemma3:4b` — older naming, replaced 2026-05-12\n- ❌ `gemma3n:e4b` — wrong family, replaced 2026-05-12\n- ✅ `gemma4:e4b` — correct (per user's Ollama installation screenshot)\n\nIf unsure, ask the user to run `ollama list` and verify.\n\n## Canonical Anthropic models\n\n| Model | Use case | Pricing (per 1M tokens) |\n|---|---|---|\n| **`claude-fable-5-1`** | Highest widely released Claude tier; 1M context, 128K max output, and stronger long-running agentic work | $10 input / $50 output; $0.25 cache read |\n| **`claude-mythos-5-1`** | Same model as Fable 5.1, with access limited to vetted cybersecurity and life-science users | $10 input / $50 output; $0.25 cache read |\n| **`claude-haiku-5-5`** | Current small-model tier for scoped tasks; 1M context / 128K output. Stage 1 exercises 1 and 3 use 5.5; other exercises retain Haiku 4.5 compatibility defaults pending parameter, thinking-block, and token-budget migration | Prompts ≤100K: $0.10 input / $0.50 output; >100K: $0.50 / $2.50 |\n| **`claude-haiku-4-5`** | Existing exercise compatibility baseline, not the latest or cheapest Haiku | $1 input / $5 output |\n| **`claude-sonnet-5-5`** | Production default for new work; review the [Sonnet 5.5 migration guide](https://platform.claude.com/docs/en/models/sonnet-5-5/migration-guide) before changing existing tool-calling examples | $2 input / $10 output; $0.10 cache read |\n| **`claude-opus-5-5`** | Current Opus-class default for most workloads; use Fable 5.1 when evals still fall short | $4 input / $20 output; $0.20 cache read |\n\n## Framing rules (do not violate)\n\n1. **Claude is the canonical / production reference** in documentation positioning.\n2. **Ollama is the practice default** because of cost — students should not be blocked by API fees during learning.\n3. **Every exercise must ship BOTH paths**:\n   - Path A (Ollama, primary practice runnable): keep the exercise title, result,\n     and first action visible. Use `<details markdown=\"1\" open>` only when Path A\n     is the single immediate action and its rendered body is short; otherwise use\n     a closed `<details markdown=\"1\">` block for code and troubleshooting.\n   - Path B (Anthropic, `<details markdown=\"1\">`, optional cloud-quality comparison)\n4. **Every exercise must mention budget explicitly** — single-run cost + total stage cost.\n5. **Local LLMs must appear in any model recommendation list** — never list cloud-only options.\n\n## Exercise file conventions\n\n- `starter.py` = Ollama / OpenAI-compatible default (Path A)\n- `starter_anthropic.py` = Anthropic SDK version (Path B)\n- `test.py` = mock-based tests for the Ollama starter (OpenAI-compat response shape)\n- `test_anthropic.py` = mock-based tests for the Anthropic starter (content-block shape)\n- `requirements.txt` = both `openai` and `anthropic` pinned\n- `README.md` = trilingual switcher + 怎麼跑（兩條 path）+ budget per path + walkthrough + common pitfalls\n- Each starter ends with `# === 自我驗證 ===` block containing 2+ `assert` statements\n- Each Python file headers Windows-cp950 UTF-8 reconfigure:\n  ```python\n  import sys\n  if hasattr(sys.stdout, \"reconfigure\"):\n      sys.stdout.reconfigure(encoding=\"utf-8\", errors=\"replace\")\n  ```\n\n## Translation rules\n\n- **zh-TW canonical** (`.md` without language suffix). zh-Hans + en mirror.\n- Freeze the Traditional Chinese meaning before translation. A bounded translation agent may produce en + zh-Hans only after its file scope, URLs, numbers, headings, and safety boundaries are fixed.\n- Mechanical conversion is only a first pass. Run the Hans, mirror, anchor, and locale-link gates, then do a human-readable semantic comparison.\n\n## Codex delegation rules\n\n- The primary agent owns scope, architecture, governance, final integration, Git, and user communication.\n- Delegated executors receive bounded file ownership, acceptance commands, a return contract, and a stop condition. They must not revert concurrent work.\n- A separate reviewer reads the final stable staged diff. Any later edit invalidates that review fingerprint.\n- Agent boundaries do not create commit boundaries. Stage explicit paths only, run the required gates, and commit the accepted integrated result.\n\n## Current curriculum contract (verified 2026-09-13)\n\n| Component | Status |\n|---|---|\n| Public curriculum | Stage 0–8, Stage 7.5, A1–A3, five role paths, walkthrough, Capstone, Glossary, and core resource pages have Traditional Chinese, Simplified Chinese, and English routes. |\n| Reader path | Enrolled pages keep goals, bold core terms, required reading, rated projects/resources, exercise outcomes, and completion checks visible; setup, long code, alternatives, and troubleshooting may be collapsed. |\n| Examples | Model-backed teaching folders keep a free/local Ollama path, an optional Anthropic path, budget guidance, and offline behavior tests unless the exercise is deliberately model-free. |\n| Stage 5 | The chapter contains five cumulative exercises plus 5.1–5.8 reference entrances; the tool-calling tutor remains the installable meta-example. |\n| Stage 6 | The reader path, advanced RAG/Memory pages, isolated collections, chunk-overlap guard, persistent memory, and offline behavior tests are present. Live model output quality is not claimed. |\n| Stage 7 | The main order is Eval → Observability → Approval/Recovery → Deploy; Multi-Agent remains optional. Six example folders cover the production mechanics. |\n| Automated checks | On 2026-09-13, 58 `scripts/test_*.py` modules collect 1,145 tests. Counts are a dated observation; CI and `pytest --collect-only` are the current source of truth. |\n| Merge gate | `Required / pr-gate` is the stable required check. A green machine gate does not replace maintainer review. |\n\n<!-- File: .claude-plugin/marketplace.json (Tokens: ~300 | Category: marketplace) -->\n# Claude Plugin Marketplace Catalog\n\nPath: `.claude-plugin/marketplace.json`\n\n[View Raw Content on GitHub](https://raw.githubusercontent.com/WenyuChiou/awesome-agentic-ai-zh/HEAD/.claude-plugin/marketplace.json)\n\n<!-- File: .claude-plugin/plugin.json (Tokens: ~300 | Category: plugin-manifest) -->\n# claude-plugin Plugin Manifest\n\nPath: `.claude-plugin/plugin.json`\n\n[View Raw Content on GitHub](https://raw.githubusercontent.com/WenyuChiou/awesome-agentic-ai-zh/HEAD/.claude-plugin/plugin.json)\n\n<!-- File: .cursor/mcp.json (Tokens: ~300 | Category: mcp-config) -->\n# Model Context Protocol (MCP) Configuration\n\nPath: `.cursor/mcp.json`\n\n[View Raw Content on GitHub](https://raw.githubusercontent.com/WenyuChiou/awesome-agentic-ai-zh/HEAD/.cursor/mcp.json)\n</yakaai_skills>"}