{"owner":"skyzh","repo":"tiny-llm","hasSkills":true,"hasMcp":false,"mcpConfig":null,"found":["AGENTS.md"],"skills":{"AGENTS.md":"# AGENTS.md\n\n## Scope\n\n- This file applies to the entire repository.\n- Use this as the default test-running policy for coding agents.\n\n## Objective\n\n- Run and verify tests in a way that matches the book workflow (`book/src/*.md`).\n- Prefer `pdm` entrypoints defined in `pyproject.toml`.\n\n## Environment Requirements\n\n- macOS on Apple Silicon is expected by the project.\n- Install dependencies first:\n\n```bash\npdm install -v\npdm run check-installation\n```\n\n- Optional baseline check from the setup chapter (reference solution, Week 1):\n\n```bash\npdm run test-refsol -- -- -k week_1\n```\n\n## Agent Test Workflow\n\n1. Start with the smallest relevant scope (`--week` + `--day`).\n2. Use pytest filters via `-- -k ...` to isolate failing tasks.\n3. Run broader suites only after targeted tests pass.\n4. If extension code changed, rebuild extensions before testing.\n\n## Canonical Commands\n\nRun all tests:\n\n```bash\npdm run test\n```\n\nRun a specific chapter/day:\n\n```bash\npdm run test --week <WEEK> --day <DAY>\n```\n\nRun with pytest filters:\n\n```bash\npdm run test --week 1 --day 3 -- -k task_2\npdm run test --week 2 --day 3 -- -k task_3\npdm run test --week 2 --day 3 -- -k gpu\n```\n\nRun reference-solution tests:\n\n```bash\npdm run test-refsol\npdm run test-refsol --week 2 --day 3 -- -k task_3\n```\n\n## Extension Rebuild Rule\n\nRebuild before tests if these changed:\n\n- `src/extensions/src/*`\n\nCommands:\n\n```bash\npdm run build-ext\n```\n\n## Guardrails\n\n- Use `--` before pytest args (`-k`, `-q`, `--collect-only`, etc.).\n- `pdm run test --week X --day Y` auto-copies `tests_refsol/test_week_X_day_Y.py` into `tests/`.\n- Model-dependent tests (0.5B/1.5B/7B) skip when models are not downloaded locally.\n## GitHub CLI\n\n- The `gh` CLI is installed, authenticated, and expected to work in this\n  repository.\n- If a `gh` command fails because of sandbox or network restrictions, retry it\n  with the required authorization. Do not infer that GitHub or `gh` is\n  unavailable from a sandboxed failure.\n"},"files":{"AGENTS.md":"# AGENTS.md\n\n## Scope\n\n- This file applies to the entire repository.\n- Use this as the default test-running policy for coding agents.\n\n## Objective\n\n- Run and verify tests in a way that matches the book workflow (`book/src/*.md`).\n- Prefer `pdm` entrypoints defined in `pyproject.toml`.\n\n## Environment Requirements\n\n- macOS on Apple Silicon is expected by the project.\n- Install dependencies first:\n\n```bash\npdm install -v\npdm run check-installation\n```\n\n- Optional baseline check from the setup chapter (reference solution, Week 1):\n\n```bash\npdm run test-refsol -- -- -k week_1\n```\n\n## Agent Test Workflow\n\n1. Start with the smallest relevant scope (`--week` + `--day`).\n2. Use pytest filters via `-- -k ...` to isolate failing tasks.\n3. Run broader suites only after targeted tests pass.\n4. If extension code changed, rebuild extensions before testing.\n\n## Canonical Commands\n\nRun all tests:\n\n```bash\npdm run test\n```\n\nRun a specific chapter/day:\n\n```bash\npdm run test --week <WEEK> --day <DAY>\n```\n\nRun with pytest filters:\n\n```bash\npdm run test --week 1 --day 3 -- -k task_2\npdm run test --week 2 --day 3 -- -k task_3\npdm run test --week 2 --day 3 -- -k gpu\n```\n\nRun reference-solution tests:\n\n```bash\npdm run test-refsol\npdm run test-refsol --week 2 --day 3 -- -k task_3\n```\n\n## Extension Rebuild Rule\n\nRebuild before tests if these changed:\n\n- `src/extensions/src/*`\n\nCommands:\n\n```bash\npdm run build-ext\n```\n\n## Guardrails\n\n- Use `--` before pytest args (`-k`, `-q`, `--collect-only`, etc.).\n- `pdm run test --week X --day Y` auto-copies `tests_refsol/test_week_X_day_Y.py` into `tests/`.\n- Model-dependent tests (0.5B/1.5B/7B) skip when models are not downloaded locally.\n## GitHub CLI\n\n- The `gh` CLI is installed, authenticated, and expected to work in this\n  repository.\n- If a `gh` command fails because of sandbox or network restrictions, retry it\n  with the required authorization. Do not infer that GitHub or `gh` is\n  unavailable from a sandboxed failure.\n"},"items":[{"name":"AGENTS.md","path":"AGENTS.md","title":"AGENTS.md","content":"# AGENTS.md\n\n## Scope\n\n- This file applies to the entire repository.\n- Use this as the default test-running policy for coding agents.\n\n## Objective\n\n- Run and verify tests in a way that matches the book workflow (`book/src/*.md`).\n- Prefer `pdm` entrypoints defined in `pyproject.toml`.\n\n## Environment Requirements\n\n- macOS on Apple Silicon is expected by the project.\n- Install dependencies first:\n\n```bash\npdm install -v\npdm run check-installation\n```\n\n- Optional baseline check from the setup chapter (reference solution, Week 1):\n\n```bash\npdm run test-refsol -- -- -k week_1\n```\n\n## Agent Test Workflow\n\n1. Start with the smallest relevant scope (`--week` + `--day`).\n2. Use pytest filters via `-- -k ...` to isolate failing tasks.\n3. Run broader suites only after targeted tests pass.\n4. If extension code changed, rebuild extensions before testing.\n\n## Canonical Commands\n\nRun all tests:\n\n```bash\npdm run test\n```\n\nRun a specific chapter/day:\n\n```bash\npdm run test --week <WEEK> --day <DAY>\n```\n\nRun with pytest filters:\n\n```bash\npdm run test --week 1 --day 3 -- -k task_2\npdm run test --week 2 --day 3 -- -k task_3\npdm run test --week 2 --day 3 -- -k gpu\n```\n\nRun reference-solution tests:\n\n```bash\npdm run test-refsol\npdm run test-refsol --week 2 --day 3 -- -k task_3\n```\n\n## Extension Rebuild Rule\n\nRebuild before tests if these changed:\n\n- `src/extensions/src/*`\n\nCommands:\n\n```bash\npdm run build-ext\n```\n\n## Guardrails\n\n- Use `--` before pytest args (`-k`, `-q`, `--collect-only`, etc.).\n- `pdm run test --week X --day Y` auto-copies `tests_refsol/test_week_X_day_Y.py` into `tests/`.\n- Model-dependent tests (0.5B/1.5B/7B) skip when models are not downloaded locally.\n## GitHub CLI\n\n- The `gh` CLI is installed, authenticated, and expected to work in this\n  repository.\n- If a `gh` command fails because of sandbox or network restrictions, retry it\n  with the required authorization. Do not infer that GitHub or `gh` is\n  unavailable from a sandboxed failure.\n","category":"root","tokens":494}]}