feat(tools): add get_my_profile, search_companies, get_company_employees

[Hovo-Dev]

Summary

• Adds get_my_profile to let agents fetch the authenticated user's own
  profile without knowing their username — navigates to /in/me/ which
  LinkedIn redirects to the real profile; resolved URL reveals the actual
  username. Supports the same sections and max_scrolls parameters as
  get_person_profile.
• Adds search_companies to search LinkedIn companies by keyword, directly
  mirroring the existing search_people pattern and return format.
• Adds get_company_employees to list employees from
  /company/{name}/people/ with an optional keywords filter by name,
  title, or skill — useful for "who do I know at X?" workflows. Returns
  /in/ references for each listed employee.

Implementation notes

All three tools follow existing conventions exactly:

• extract_page + _RATE_LIMITED_MSG sentinel pattern
• AuthenticationError double-catch with handle_auth_error
• readOnlyHint / openWorldHint annotations
• exclude_args=["extractor"] for test injection
• Extractor methods added to scraping/extractor.py, tools registered in
  tools/person.py and tools/company.py

Checklist

• Extractor methods added (scraping/extractor.py)
• Tools registered (tools/person.py, tools/company.py)
• Mock methods added to _make_mock_extractor (tests/test_tools.py)
• Test classes added (11 new tests, all passing)
• TestToolTimeouts updated with new tool names
• README.md tool table updated
• docs/docker-hub.md features list updated
• manifest.json tools array updated
• uv run pytest — 41 passed
• uv run ruff check . && uv run ruff format --check . — clean
• uv run ty check — clean

Synthetic prompt

Add three new MCP tools — get_my_profile (authenticated user's own
profile via /in/me/), search_companies (company keyword search
mirroring search_people), and get_company_employees (employee list
from /company/{name}/people/ with optional keyword filter). Follow the
existing extract_page + _RATE_LIMITED_MSG pattern, add extractor
methods, register tools, wire tests with mock injection, and update
README, docker-hub docs, and manifest per CONTRIBUTING.md checklist.

[greptile-apps]

Greptile Summary

This PR adds three new MCP tools — get_my_profile, search_companies, and get_company_employees — all following the existing extract_page + _RATE_LIMITED_MSG sentinel pattern, with correct AuthenticationError double-catch, readOnlyHint/openWorldHint annotations, exclude_args=[\"extractor\"] for test injection, and full README/manifest/docker-hub updates.

get_my_profile addresses the previously resolved URL-resolution concern by first navigating to /in/me/ to capture the post-redirect URL, extracting the real username via regex, and delegating to scrape_person — an improvement over always returning /in/me/. search_companies mirrors search_people exactly, and get_company_employees correctly uses quote_plus for the keyword query parameter.

Confidence Score: 5/5

Safe to merge — all three tools follow established conventions with no logic errors, correct error handling, and comprehensive test coverage.

No P0 or P1 findings. All new tools mirror existing patterns (extract_page + _RATE_LIMITED_MSG, AuthenticationError double-catch, readOnlyHint/openWorldHint, exclude_args injection). Tests cover success, sections, callbacks, unknown sections, and error paths. Docs and manifest are fully updated.

No files require special attention.

Important Files Changed

Filename	Overview
linkedin_mcp_server/scraping/extractor.py	Adds get_my_profile, get_company_employees, and search_companies extractor methods; all follow existing patterns and correctly handle rate-limit sentinel, errors, and references
linkedin_mcp_server/tools/company.py	Registers search_companies and get_company_employees tools; correct error handling, progress reporting, and tool annotations
linkedin_mcp_server/tools/person.py	Registers get_my_profile tool; correctly uses parse_person_sections, passes callbacks, and handles unknown section names
tests/test_tools.py	Adds 11 new tests covering success, section selection, callbacks, unknown sections, and error paths for all three new tools
manifest.json	Adds manifest entries for get_my_profile, search_companies, and get_company_employees

Sequence Diagram

sequenceDiagram
    participant Agent
    participant Tool as MCP Tool Layer
    participant Extractor as LinkedInExtractor
    participant Browser as Playwright
    participant LI as LinkedIn

    Note over Agent,LI: get_my_profile flow
    Agent->>Tool: get_my_profile(sections)
    Tool->>Extractor: get_my_profile(sections, callbacks)
    Extractor->>Browser: navigate /in/me/
    Browser->>LI: GET /in/me/
    LI-->>Browser: redirect to real profile
    Browser-->>Extractor: page.url = real profile URL
    Extractor->>Extractor: regex extract real username
    Extractor->>Browser: scrape_person with real username
    Browser->>LI: GET each section URL
    LI-->>Browser: page content
    Extractor-->>Agent: url + sections dict

    Note over Agent,LI: search_companies flow
    Agent->>Tool: search_companies(keywords)
    Tool->>Extractor: search_companies(keywords)
    Extractor->>Browser: extract_page search URL
    Browser->>LI: GET company search
    LI-->>Browser: results page
    Extractor-->>Agent: url + search_results section

    Note over Agent,LI: get_company_employees flow
    Agent->>Tool: get_company_employees(company, kw)
    Tool->>Extractor: get_company_employees(company, kw)
    Extractor->>Browser: extract_page /company/name/people/
    Browser->>LI: GET employees page
    LI-->>Browser: employees list
    Extractor-->>Agent: url + employees section + references

Loading

_{Reviews (2): Last reviewed commit: "fix(get_my_profile): resolve real userna..." | Re-trigger Greptile}

[Hovo-Dev]

@stickerdaniel Thank you for building and maintaining this — it's genuinely one of the most useful MCP servers out there that I use everyday for researching through my background AI agents.

I put together a few tools that I think would be a great addition: get_my_profile to let agents know who they're acting as, search_companies to find companies by keyword, and get_company_employees to list people at a company with an optional filter. I'm actively running AI agents that research LinkedIn and I kept hitting the wall without these — having them would make a real difference both for my workflows and for others who use this on a daily basis.

Would really appreciate your review and a merge if it looks good to you!