2-repro-issue

Reproduce LinkedIn MCP server issues locally against a real authenticated session.

Updated Aug 20, 2026
One-click install
npx skills add https://github.com/abetoluwani/linkedin-mcp-server --skill 2-repro-issue-abetoluwani
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: 2-repro-issue
Source: https://github.com/abetoluwani/linkedin-mcp-server/tree/main/.agents/skills/2-repro-issue
Command: npx skills add https://github.com/abetoluwani/linkedin-mcp-server --skill 2-repro-issue-abetoluwani

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Confirming whether a reported bug in the LinkedIn MCP server actually exists requires manually starting the server, authenticating, and replaying the failing tool call. This Skill automates that reproduction workflow and produces concrete evidence tied to the exact code path. ## Core Features & Use Cases - Issue Parsing and Mapping: Extracts the affected MCP tool, arguments, and expected behavior from a GitHub issue, then maps them to the scraper code path in linkedin_mcp_server. - Live Reproduction: Starts the MCP streamable-http server from the current workspace with uv run, initializes an MCP session, and calls the failing tool against the real authenticated LinkedIn profile. - Verdict and Baseline: Classifies the result as reproduced, different mode, not reproduced, or inconclusive, and saves the response baseline for later PR-fix verification. - Use Case: A maintainer sees issue #442 claiming get_person_profile returns empty sections. They run this Skill with the issue number and receive a verdict with evidence JSON and the likely failing file. ## Quick Start Ask the assistant to reproduce issue 442 from the linkedin-mcp-server repository locally using this Skill.

Frequently Asked Questions about 2-repro-issue

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I reproduce a GitHub issue for an MCP server locally?▼

Pass the issue number or URL to this Skill. It reads the issue with gh, starts the MCP server from your workspace with uv run, initializes a session over streamable-http, and replays the exact failing tool call to capture evidence.

How to test LinkedIn MCP tools against a real account?▼

The Skill uses the authenticated browser profile stored at ~/.linkedin-mcp/profile/ and calls tools like get_person_profile or search_jobs through the local MCP server. No mocks or fixtures are used; results reflect live LinkedIn behavior.

Does this Skill fix the bug or check out a PR?▼

No. It only reproduces and reports the issue on the current branch without editing code, committing, or checking out pull requests. Verifying a candidate fix is handled separately by the 3-verify-pr-fix Skill.

Why does the MCP initialized notification return an invalid parameters error?▼

The notifications/initialized post often returns a -32602 error with a pydantic validation dump in the server log. This is harmless; the session remains valid and tools/call works on the same Mcp-Session-Id.

What happens if the LinkedIn session profile is missing or stale?▼

The Skill checks for the profile directory before starting the server. If it is missing or stale, it runs the login flow once per invocation, and if login or browser errors appear in the log it stops and reports rather than retrying.