AgentSkillsCN

velocity-assessment

评估开发团队的执行速度与工作质量

SKILL.md
--- frontmatter
name: velocity-assessment
description: Executive velocity and quality assessment for development teams
allowed-tools: Read, Bash, Grep, Glob, Task
context: fork

Executive Velocity Assessment

Assess development velocity and code quality for executive reporting.

Usage

code
/velocity:velocity-assessment [days=90] [branch=main]

Arguments:

  • days - Analysis window (default: 90 days)
  • branch - Target branch (default: main)

Three-Agent Workflow

This skill orchestrates three specialized agents sequentially:

PhaseAgentPurposeModel
1velocity-git-analystGit commit/contributor analysissonnet
2velocity-quality-assessorCode quality via Codanna CLIsonnet
3velocity-reporterExecutive report compilationopus

Orchestration Instructions

Phase 0: Codanna Preflight Check

Before launching agents, verify Codanna is initialized:

bash
codanna config

If this returns configuration data, Codanna is initialized. Then check index status:

bash
codanna mcp get_index_info --json

If Codanna is not initialized or index is empty:

  • Set $codanna_available = false
  • Skip Phase 2 (quality assessment)
  • Note in final report: "Code quality assessment skipped - Codanna index not available"

If Codanna is ready:

  • Set $codanna_available = true
  • Proceed with all phases

Phase 1: Git Analysis

Launch the velocity-git-analyst agent with:

code
Analyze git history for the past {days} days on branch {branch}.
Output JSON with velocity metrics, contributor stats, and temporal patterns.

Wait for completion. Store output as $git_analysis.

Phase 2: Quality Assessment (Conditional)

Skip if $codanna_available = false.

If Codanna is available, launch the velocity-quality-assessor agent with:

code
Assess code quality using Codanna CLI (codanna mcp <tool> --json).
Output JSON with quality indicators and recommendations.

Wait for completion. Store output as $quality_assessment.

If skipped, set $quality_assessment = null and proceed to Phase 3.

Phase 3: Report Compilation

Launch the velocity-reporter agent (model: opus) with:

code
Compile executive velocity report from:
- Git analysis: {$git_analysis}
- Quality assessment: {$quality_assessment}

Follow the report template in metrics-definitions.md and report-template.md.

Output

Present the final executive report to the user.

Reference Files

  • metrics-definitions.md - Metric formulas and thresholds
  • report-template.md - Executive report format

Example Output

code
# Development Velocity Report

**Period:** 2025-11-03 to 2026-02-03 (90 days)
**Branch:** main

## Executive Summary

### Dashboard

| Metric | Value | Benchmark | Delta | Status |
|--------|-------|-----------|-------|--------|
| Commits/Day | 2.3 | 2.0 | +0.3 (15% over) | 🟢 Healthy |
| Bus Factor | 2 | 3 | -1 (33% under) | 🟡 Warning |
| Code Quality | 78% | 80% | -2pp (3% under) | 🟡 Warning |

## Contributor Breakdown

| Contributor | Commits | % of Total | Cumulative % | Role |
|-------------|---------|------------|--------------|------|
| Alice | 95 | 45.9% | 45.9% | Primary |
| Bob | 62 | 30.0% | 75.9% | Primary |
| Carol | 30 | 14.5% | 90.4% | Secondary |
| Dave | 15 | 7.2% | 97.6% | Peripheral |
| Eve | 5 | 2.4% | 100.0% | Peripheral |

### Contributor Analysis

Top 2 contributors (Alice, Bob) account for 75.9% of commits. With a bus factor of 2, the project has emerging key-person risk. Ideal distribution for 5 contributors is 20% each. Alice exceeds ideal by 26 percentage points. Knowledge transfer to Carol recommended.

## Key Findings

1. Commits/day at 2.3 exceeds benchmark by 15%
2. Bus factor of 2 falls 33% below target of 3
3. Error handling coverage at 85% exceeds threshold

## Recommendations

1. Pair programming sessions between Alice and Carol
2. Address complexity hotspots in parser module