llms.txt

# FlyPython Repository Index for LLMs and AI Agents

> FlyPython provides production-tested Python engineering playbooks, challenge courses with objective verification and per-checkpoint claim codes, verifiable test-driven starter examples, task contracts, and agent IDE rules.

## Quick API and Automation Commands

- Install development dependencies: `python -m pip install -r requirements-dev.lock.txt`
- Run every CI check (lint, tests, exports, examples, courses, paths): `make check`
- Run the test suite only: `python -m pytest`
- Verify runnable examples: `python tools/verify_examples.py`
- Verify course folders: `python tools/verify_courses.py`
- Validate and inspect resource catalog: `python tools/validate_catalog.py`
- Discover Radar review candidates: `python tools/radar_scan.py --dry-run`
- Structured Content Manifest: `content-manifest.json`
- Full curated catalog data: `catalog.json`
- Project Radar data: `radar.json`

## Machine-Readable Directories

- `catalog/resources/`: YAML files representing verified libraries, tools, and platforms.
- `catalog/projects/`: One reviewed YAML file per Project Radar entry (discovery candidates are produced on demand by `tools/radar_scan.py`).
- `courses/`: Challenge course folders — solve the `TASK.md` contract with a coding agent and prove it with `verify.py` claim codes; `COURSE.md` still offers a guided, agent-taught mode:
  - `courses/hands-on-python-with-claude-code/`: Task contract, bounded change, and objective verify.py evidence with Claude Code (EN+ZH lessons, three scenario skins).
  - `courses/hands-on-with-openai-codex/`: The same verified workflow taught hands-on in the Codex desktop app, where an agent thread runs the loop and you supervise diffs and approvals (EN+ZH).
  - `courses/hands-on-with-cursor/`: The same verified workflow taught hands-on in Cursor, where Agent mode runs the loop and you supervise diffs and project rules (EN+ZH).
  - `courses/hands-on-with-deepseek-harness/`: The same verified workflow taught hands-on in DeepSeek Harness, where the agent loop is composed from plugins and every step lands in an inspectable trajectory (EN+ZH).
  - `courses/hands-on-with-kimi-code/`: The same verified workflow taught hands-on in Kimi Code, where built-in subagents split read-only exploration, planning, and the edits you approve (EN+ZH).
  - `courses/hands-on-with-zcode/`: The same verified workflow taught hands-on in the ZCode desktop workspace, where Goal Mode decomposes the work and the Git state shows every change (EN+ZH).
  - `courses/hands-on-with-github-copilot/`: The same verified workflow taught hands-on with GitHub Copilot in VS Code, where the Agent role runs the loop and you supervise permissions, instructions files, and diffs (EN+ZH).
  - `courses/agent-rules-single-source/`: Build a checker that proves one AGENTS.md source of truth across CLAUDE.md and .cursorrules (EN+ZH, three drift scenarios).
  - `courses/verifying-ai-generated-code/`: Build the release-evidence machine: run checks, parse results, refuse zero-test projects, write SHIP-RECORD.json (EN+ZH, three project skins).
  - `courses/llm-features-in-python/`: Turn a demo-day model call into a shippable feature: validate structured output, allowlist tool calls, retry with the error fed back, fall back to a human, and measure with an offline eval that divides by every case (EN+ZH, three skins).
  - `courses/secure-ai-generated-code/`: Review a plausible AI-style notes service like an attacker: close SQL injection, path traversal, a hard-coded secret, fast password hashes, and pickled sessions, then replay the attacks (EN+ZH, two skins).
  - `courses/mcp-server-in-python/`: Build a stateless 2026-07-28 MCP tool server in pure Python — reuses the reviewed mcp-server example core (EN+ZH).
  - `courses/da-eda/`: Exploratory data analysis with pandas over a messy shop export: exact duplicates, bad amounts, one bad date (EN+ZH).
  - `courses/da-visualization/`: Data visualization with matplotlib from clean long-format daily sales: summaries, one-figure charts, Agg-before-pyplot discipline (EN+ZH).
  - `courses/da-report/`: Turn a verified EDA summary JSON into a rendered report with stdlib-only code and deterministic verification (EN+ZH).
- `examples/`: Reproducible test projects with failing starters and verified solutions:
  - `examples/product-slug/`: String normalization edge cases and regex safety.
  - `examples/pydantic-validation/`: Model boundaries and custom validators.
  - `examples/mcp-server/`: JSON-RPC 2.0 Model Context Protocol tool server with schema validation.
  - `examples/async-fetcher/`: Async concurrency control with Semaphore and backoff retry.
  - `examples/structured-pipeline/`: Batch data pipeline with error isolation and schema checking.
- `templates/`:
  - `templates/AGENT_RULES.example.md`: Universal rule set for Cursor (.cursorrules), Windsurf, and Copilot.
  - `templates/AGENT_RULES.example_cn.md`: Universal rule set in Chinese.
  - `templates/AGENTS.example.md`: Repository instruction policy for coding agents.
  - `templates/AGENTS.example_cn.md`: Repository instruction policy in Chinese.
  - `templates/TASK_CONTRACT.md`: Scoped specification template for coding agents.
  - `templates/IMPLEMENTATION_PLAN.md`: Implementation plan template.
  - `templates/CODE_REVIEW.md`: Code change review template.
  - `templates/VERIFICATION.md`: Evidence record for reproducible tests.

## Official Portal

- Official Website: https://flypython.com
- Website LLMs Index: https://flypython.com/llms.txt
- Complete Text Knowledge Dump: https://flypython.com/llms-full.txt