git2linkedin（git2linkedin 命令行）

Git2LinkedIn

Overview

Generate a LinkedIn Experience draft from git history by directly reading commit messages and converting them into feature-focused, business-relevant, and confidentiality-safe bullets.

Workflow

1. Collect required inputs:

• role (required)
• company (required)

1. Collect optional inputs:

• repo (default .)
• since (YYYY-MM-DD)
• until (YYYY-MM-DD)
• author (default: current git user)
• out (optional markdown output path)
• max-commits (default 400)

1. Run the extractor:

# All-time
python3 skills/git2linkedin/scripts/git_history_extract.py \
  --role "Senior Software Engineer" \
  --company "Acme"

# Date range
python3 skills/git2linkedin/scripts/git_history_extract.py \
  --role "Senior Software Engineer" \
  --company "Acme" \
  --since 2025-01-01 \
  --until 2025-12-31

1. Convert JSON output into final LinkedIn text:

• Write one short English summary sentence.
• Write 4-6 impact bullets.
• Focus on end-user value and business relevance.
• Avoid repository metrics, commit hashes, file paths, and internal identifiers.

1. Handle empty ranges explicitly:

• If commit list is empty, explain that no commits were found and suggest widening the date range or changing author.

Output Contract

Use these JSON fields:

• summary: baseline sentence for LinkedIn experience
• feature_highlights: feature-oriented bullet candidates
• end_user_outcomes: user-facing outcomes
• business_relevance: why this work matters for role/business goals
• source_mode: extraction mode metadata (direct-commit-messages)
• guidance: fallback suggestion when history is empty

Guardrails

• Keep output in English unless user asks otherwise.
• Keep bullets concise and outcome-focused.
• Keep sentence style mostly passive and neutral; avoid first-person phrasing (I improved, I enhanced, etc.).
• Do not include repository statistics.
• Do not expose commit hashes, file paths, issue IDs, URLs, or raw commit messages.
• Do not invent confidential product details; keep language high-level and safe for public profiles.