name: contributor-analyzer
description: >
Analyzes engineering contributors for annual reviews, progress tracking,
and production ownership. Reads every commit diff (not sampling), calculates
accuracy rates, assesses engineering maturity through qualitative judgment,
and generates promotion-ready performance reviews. Works with any git
repository using local git for commit analysis and gh CLI for GitHub
PR/review metadata.
license: MIT
compatibility: "Requires git (local repository access) and gh (GitHub CLI) for full analysis. Optional: jq for JSON parsing, bc for arithmetic."
metadata:
author: anivar
version: 1.0.0
tags:
- contributor-analysis
- annual-review
- code-quality
- production-ownership
- promotion
- engineering-maturity

Contributor Analyzer

Unified skill for engineering contributor analysis. Combines annual review generation, historical progress tracking, production ownership mapping, and qualitative judgment — all driven by reading actual commit diffs.

Skill Format

Each reference file covers one analysis concern:

• Quick Reference: Summary tables and key thresholds
• Process Steps: Bash commands and orchestration
• Templates: Output formats for reports and profiles
• Deep Dive: Full context with examples from real-world 1,091-commit analysis

Impact ratings: CRITICAL (blocks accuracy), HIGH (major insight), MEDIUM (enriches analysis)

When to Apply

Reference these guidelines when:
- Running annual or quarterly performance reviews
- Assessing promotion readiness with evidence
- Tracking contributor progress over time (incremental updates)
- Mapping production code ownership and risk
- Comparing multiple engineers for team decisions
- Identifying single points of failure in codebase ownership

Priority-Ordered References

Quick Reference

Critical: Context Management

Before launching ANY analysis, count commits:

git log --author="EMAIL" --after="YEAR-01-01" --before="YEAR+1-01-01" --oneline | wc -l

Batch sizing (hard limits from real failures):

Sub-tasks write to files, return 3-line summaries. Never return raw analysis inline.

Critical: Annual Review Process

7 phases, sequential:
1. Identity Discovery — find all git email variants
2. Metrics — commits, PRs, reviews, lines (git + gh CLI)
3. Read ALL Diffs — quarterly parallel sub-tasks, file-based output
4. Bug Introduction — self-reverts, crash-fixes, same-day fixes, hook bypass
5. Code Quality — anti-patterns and strengths from diff reading
6. Report Generation — structured markdown with rating + promotion assessment
7. Comparison — multi-engineer ranking with evidence

High: Accuracy Rate

Effective Accuracy = 100% - (fix-related commits / total commits)

High: Tool Separation

• Local git: ALL commit-level analysis (log, show, diff, blame, shortlog)
• gh CLI: ONLY for PR counts, review counts, user verification
• Never use gh to read diffs — local git is faster with no rate limits

References

Full documentation with process steps and templates in references/:

Analysis Process

Judgment & Assessment

Ownership & Progress

Searching References

# Find by keyword
grep -rl "accuracy" references/
grep -rl "promotion" references/
grep -rl "ownership" references/
grep -rl "anti-pattern" references/

Problem to Reference Mapping

Full Compiled Document

For the complete guide with all references expanded: AGENTS.md

Usage Examples

# Annual review (single engineer)
Analyze @alice for 2025 annual review in repo org/repo.
Git email: [email protected]. Write to ./alice-2025-review.md

# Multi-engineer comparison with promotion decision
Analyze @alice, @bob, @charlie for 2025 reviews.
I need to decide which 2 get promoted.

# Production ownership mapping
Analyze production code ownership in this repo.

# Incremental progress update
Update @alice's progress profile with latest contributions.

# Critical path risk detection
Find single points of failure in production code ownership.

# Plateau detection
Has @bob plateaued? Check last 3 quarters.