Files

T

Pnant 380cfd565d feat(identity): capability-layer positioning + skill rewrite, v1.5.0

SKILL.md (per the skill-writing guide):
- description rewritten trigger-first: 调研/全网调研/帮我调研/research lead
  the text (the owner's own phrasing "去全网调研" previously had NO matching
  trigger word — the exact undertrigger the guide warns about); platform
  keywords front-loaded against the 1536-char truncation
- standing rules added: probe `doctor --json` active_backend before acting
  on multi-backend platforms, announce which skill/backend is in use,
  follow the documented retry chains, compose multi-platform research
- quick commands updated to the new reality: bili-cli for bilibili search,
  Reddit/xiaohongshu moved to a "login-backed platforms" section
- SKILL_en.md: same description surgery + the actively harmful stale
  advice removed (anonymous reddit .json curl, yt-dlp for bilibili)

README / README_en:
- tagline gains the capability-layer subline: backends are chosen,
  installed and health-checked for us to swap — the user never notices
- "设计理念" reframed from scaffolding to capability layer (selection /
  install / doctor / routing); channels diagram now shows ordered backend
  lists; selection table split into primary + fallback columns with the
  live-test rationale per row
- platform table: honest rows for bilibili (bili-cli, yt-dlp retired),
  Reddit (no zero-config path), xiaohongshu (OpenCLI / mcp / legacy)

Version 1.5.0 across pyproject / __init__ / CLAUDE.md / test fixture.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

2026-06-11 16:30:56 +08:00

2.2 KiB

Raw Permalink Blame History

CLAUDE.md

Project

Agent Reach — Python CLI + library that gives AI agents read/search access to 13 internet platforms. Positioning: installer + doctor + config tool. NOT a wrapper — after install, agents call upstream tools directly. Repo: github.com/Panniantong/Agent-Reach | License: MIT | Version: 1.5.0

Commands

pip install -e . — Dev install
pytest tests/ -v — All tests
pytest tests/test_cli.py -v — CLI tests only
bash test.sh — Full integration test (creates venv, installs, runs doctor + channel tests)
python -m agent_reach.cli doctor — Run diagnostics
python -m agent_reach.cli install --env=auto — Auto-configure

Structure

agent_reach/cli.py — CLI entry point (argparse)
agent_reach/core.py — Core read/search routing logic
agent_reach/config.py — Config management (YAML, env vars)
agent_reach/doctor.py — Diagnostics engine
agent_reach/channels/ — One file per platform (twitter.py, reddit.py, youtube.py, etc.)
agent_reach/channels/base.py — Base channel class (all channels inherit from this)
agent_reach/integrations/mcp_server.py — MCP server integration
agent_reach/skill/ — OpenClaw skill files
agent_reach/guides/ — Usage guides
tests/ — pytest tests
config/mcporter.json — MCP tool config

Conventions

Python 3.10+ with type hints
Each channel is a single file in channels/, inherits from BaseChannel
Channel contract: must implement can_handle(url), read(url), search(query), check() methods
Use loguru for logging, rich for CLI output
Commit format: type(scope): message (one commit = one thing)
All upstream tool calls go through public API/CLI, never hack internals

Rules

NEVER modify upstream open source projects' source code
Agent Reach is a "glue layer" — only route and call, don't reimagine
Version in THREE places must match: pyproject.toml, __init__.py, tests/test_cli.py
Always new branch for changes, PR to main, never push to main directly
Run pytest tests/ -v before committing — all tests must pass
Cookie-based auth (Twitter, XHS): use Cookie-Editor export method only, no QR scan
XHS login: Cookie-Editor browser export only (QR will hang)

2.2 KiB Raw Permalink Blame History