sawz/Agent-Reach
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

refactor: strip to installer + doctor + docs, remove read/search wrapper layer

Browse Source 
BREAKING CHANGE: Remove all `agent-reach read` and `agent-reach search-*` commands.

Agent Reach is now an installer, configuration tool, and doctor —
not a wrapper layer. After installation, agents call upstream tools
directly (bird CLI, yt-dlp, mcporter, gh CLI, Jina Reader, etc.).

What's kept:
- agent-reach install: one-shot installer
- agent-reach doctor: channel status overview
- agent-reach configure: cookies, proxy, credentials
- agent-reach setup: interactive wizard
- SKILL.md: complete guide for agents to use upstream tools directly

What's removed:
- agent-reach read URL (and all channel read() methods)
- agent-reach search-* commands (and all channel search() methods)
- ReadResult / SearchResult data classes
- URL routing system (get_channel_for_url)
- All parsing/conversion logic (VTT, Reddit JSON, bird text parser, etc.)
- MCP server read/search tools (kept only get_status)

Net change: -1790 lines. Less code = fewer bugs.
This commit is contained in:
Panniantong
2026-02-26 08:15:56 +01:00
parent 1cbf6a7b9c
commit a37e9aa190
18 changed files with 400 additions and 2190 deletions
Download Patch File Download Diff File Expand all files Collapse all files
agent_reach/channels/linkedin.py
+22 -251
View File
@@ -1,268 +1,39 @@
# -*- coding: utf-8 -*-
"""LinkedIn — via linkedin-scraper-mcp (MCP) or Jina Reader fallback.
Backend: linkedin-scraper-mcp (916 stars, Patchright browser automation)
Swap to: any LinkedIn access tool
"""
"""LinkedIn — check if linkedin-scraper-mcp is available."""
import shutil
import subprocess
from urllib.parse import urlparse
from .base import Channel, ReadResult, SearchResult
from typing import List
import requests
def _mcporter_has_linkedin() -> bool:
"""Check if mcporter has linkedin MCP configured."""
if not shutil.which("mcporter"):
return False
try:
r = subprocess.run(
["mcporter", "list"], capture_output=True, text=True, timeout=10
)
return "linkedin" in r.stdout.lower()
except Exception:
return False
def _mcporter_call(expr: str, timeout: int = 30) -> str:
"""Call a LinkedIn MCP tool via mcporter."""
r = subprocess.run(
["mcporter", "call", expr],
capture_output=True, text=True, timeout=timeout,
)
if r.returncode != 0:
raise RuntimeError(r.stderr or r.stdout)
return r.stdout
from .base import Channel
class LinkedInChannel(Channel):
name = "linkedin"
description = "LinkedIn 个人/公司 Profile 和职位"
description = "LinkedIn 职业社交"
backends = ["linkedin-scraper-mcp", "Jina Reader"]
tier = 2
def can_handle(self, url: str) -> bool:
domain = urlparse(url).netloc.lower()
return "linkedin.com" in domain
from urllib.parse import urlparse
return "linkedin.com" in urlparse(url).netloc.lower()
def check(self, config=None):
if _mcporter_has_linkedin():
return "ok", "完整可用（Profile、公司、职位搜索）"
# Check if linkedin-scraper-mcp is installed as CLI
if shutil.which("linkedin-scraper-mcp"):
return "warn", (
"linkedin-scraper-mcp 已安装但未接入 mcporter。运行：\n"
" 1. linkedin-scraper-mcp --login（在有浏览器的机器上登录）\n"
" 2. linkedin-scraper-mcp --transport streamable-http --port 8001\n"
" 3. mcporter config add linkedin http://localhost:8001/mcp"
if not shutil.which("mcporter"):
return "off", (
"基本内容可通过 Jina Reader 读取。完整功能需要：\n"
" pip install linkedin-scraper-mcp\n"
" mcporter config add linkedin http://localhost:3000/mcp\n"
" 详见 https://github.com/stickerdaniel/linkedin-mcp-server"
)
try:
r = subprocess.run(
["mcporter", "list"], capture_output=True, text=True, timeout=10
)
if "linkedin" in r.stdout.lower():
return "ok", "完整可用（Profile、公司、职位搜索）"
except Exception:
pass
return "off", (
"可通过 Jina Reader 读取部分内容。完整功能需要：\n"
" 1. pip install linkedin-scraper-mcp\n"
" 2. linkedin-scraper-mcp --login（在有浏览器的机器上登录）\n"
" 3. linkedin-scraper-mcp --transport streamable-http --port 8001\n"
" 4. mcporter config add linkedin http://localhost:8001/mcp\n"
" 详见 https://github.com/stickerdaniel/linkedin-mcp-server"
"mcporter 已装但 LinkedIn MCP 未配置。运行：\n"
" pip install linkedin-scraper-mcp\n"
" mcporter config add linkedin http://localhost:3000/mcp"
)
async def read(self, url: str, config=None) -> ReadResult:
path = urlparse(url).path.strip("/")
# Try MCP first
if _mcporter_has_linkedin():
try:
if "/in/" in url:
return await self._read_profile_mcp(url)
elif "/company/" in url:
return await self._read_company_mcp(url)
elif "/jobs/view/" in url:
return await self._read_job_mcp(url)
except Exception:
pass # Fall through to Jina
# Fallback: Jina Reader
return await self._read_jina(url)
async def _read_profile_mcp(self, url: str) -> ReadResult:
"""Read a LinkedIn profile via MCP."""
import re
# Extract username from URL: /in/username/
match = re.search(r"/in/([^/]+)", url)
if not match:
return await self._read_jina(url)
username = match.group(1)
safe_username = username.replace('"', '\\"')
out = _mcporter_call(
f'linkedin.get_person_profile(linkedin_username: "{safe_username}")',
timeout=60,
)
return ReadResult(
title=self._extract_title(out) or f"LinkedIn Profile - {username}",
content=out.strip(),
url=url,
platform="linkedin",
)
async def _read_company_mcp(self, url: str) -> ReadResult:
"""Read a LinkedIn company page via MCP."""
import re
# Extract company name from URL: /company/name/
match = re.search(r"/company/([^/]+)", url)
if not match:
return await self._read_jina(url)
company = match.group(1)
safe_company = company.replace('"', '\\"')
out = _mcporter_call(
f'linkedin.get_company_profile(company_name: "{safe_company}")',
timeout=60,
)
return ReadResult(
title=self._extract_title(out) or "LinkedIn Company",
content=out.strip(),
url=url,
platform="linkedin",
)
async def _read_job_mcp(self, url: str) -> ReadResult:
"""Read a LinkedIn job posting via MCP."""
import re
match = re.search(r"/jobs/view/(\d+)", url)
if not match:
return await self._read_jina(url)
job_id = match.group(1)
out = _mcporter_call(
f'linkedin.get_job_details(job_id: "{job_id}")',
timeout=30,
)
return ReadResult(
title=self._extract_title(out) or f"LinkedIn Job {job_id}",
content=out.strip(),
url=url,
platform="linkedin",
)
async def _read_jina(self, url: str) -> ReadResult:
"""Fallback: use Jina Reader."""
try:
resp = requests.get(
f"https://r.jina.ai/{url}",
headers={"Accept": "text/markdown"},
timeout=15,
)
resp.raise_for_status()
text = resp.text
# Check if content is usable
if len(text.strip()) < 100 or "Sign in" in text[:200]:
return ReadResult(
title="LinkedIn",
content=(
f"⚠️ LinkedIn 页面需要登录才能完整查看。\n\n"
f"URL: {url}\n\n"
"完整功能需安装 linkedin-scraper-mcp：\n"
" pip install linkedin-scraper-mcp\n"
" uvx linkedin-scraper-mcp --login\n"
" 详见 https://github.com/stickerdaniel/linkedin-mcp-server"
),
url=url,
platform="linkedin",
)
return ReadResult(
title=text[:100] if text else url,
content=text,
url=url,
platform="linkedin",
)
except Exception:
return ReadResult(
title="LinkedIn",
content=(
f"⚠️ 无法读取此 LinkedIn 页面: {url}\n\n"
"提示：\n"
"- LinkedIn 需要登录才能查看大部分内容\n"
"- 安装 linkedin-scraper-mcp 解锁完整功能\n"
"- 详见 https://github.com/stickerdaniel/linkedin-mcp-server"
),
url=url,
platform="linkedin",
)
async def search(self, query: str, config=None, **kwargs) -> List[SearchResult]:
limit = kwargs.get("limit", 10)
# Try MCP search first
if _mcporter_has_linkedin():
try:
return await self._search_mcp(query, limit)
except Exception:
pass
# Fallback to Exa
from agent_reach.channels.exa_search import ExaSearchChannel
exa = ExaSearchChannel()
return await exa.search(f"site:linkedin.com {query}", config=config, limit=limit)
async def _search_mcp(self, query: str, limit: int) -> List[SearchResult]:
"""Search LinkedIn via MCP."""
safe_q = query.replace('"', '\\"')
# Try job search first (most common use case)
try:
out = _mcporter_call(
f'linkedin.search_jobs(keywords: "{safe_q}")',
timeout=60,
)
results = self._parse_search_results(out, "job")
if results:
return results[:limit]
except Exception:
pass
# Try people search
try:
out = _mcporter_call(
f'linkedin.search_people(keywords: "{safe_q}")',
timeout=60,
)
results = self._parse_search_results(out, "people")
if results:
return results
except Exception:
pass
return []
def _parse_search_results(self, text: str, result_type: str) -> List[SearchResult]:
"""Parse MCP search output into SearchResults."""
import json
results = []
try:
data = json.loads(text)
items = data if isinstance(data, list) else data.get("results", data.get("jobs", []))
for item in items:
if isinstance(item, dict):
title = item.get("title") or item.get("name") or item.get("headline", "")
url = item.get("url") or item.get("link", "")
snippet = item.get("description") or item.get("company", "")
results.append(SearchResult(
title=title,
url=url,
snippet=snippet[:200] if snippet else "",
))
except (json.JSONDecodeError, KeyError):
# Try line-by-line parsing
pass
return results
def _extract_title(self, text: str) -> str:
"""Extract a title from MCP output."""
for line in text.split("\n"):
line = line.strip()
if line and not line.startswith(("{", "[", "#", "http")):
return line[:80]
return ""