Skill Analyzer — Development agent for Claude Code
Finds the patterns in eval and bench results that averages hide, assertions that never tell the skill apart, tasks that flip between runs, tools that fall behind.
How to install Skill Analyzer
Installs to ~/.claude/agents/thenavidm-skill-builder-skill-analyzer.md
mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/thenavidm/skill-builder/HEAD/agents/skill-analyzer.md -o ~/.claude/agents/thenavidm-skill-builder-skill-analyzer.md Restart Claude Code, or start a new session, for it to be picked up.
What Skill Analyzer does
name: skill-analyzer description: Finds the patterns in eval and bench results that averages hide, assertions that never tell the skill apart, tasks that flip between runs, tools that fall behind. Started by skill-eval and skill-bench.
You read a finished eval or bench (its `benchmark.json`, and each run's `grades.json`) and say what the numbers don't show at a glance.
Look for:
- Assertions that prove nothing: they pass as often without the skill as with it.
- Flaky tasks: th
Alternatives in Development
- Browser Stealth Agent — Stealth browser automation agent for targets behind Cloudflare, Akamai, Google, DataDome, or PerimeterX bot de 812 ★
- Bench Analyzer 168 ★
- Bench — API performance benchmarking — latency profiling, throughput testing, performance regression detection 73 ★
Full documentation available on GitHub
View Source RepositoryRelated Agents
Skill Grader
Grades a run's outputs against its assertions, quoting the output for every grade, so a grade without evidence
Skill Fact Checker
Lists every checkable claim a skill makes and the source behind each, or says plainly that there is none. Star
Bench Reviewer
Reviews eval suite quality — fixture/rubric consistency, scoring accuracy, false positive traps, difficulty ca
Bench Runner
Executes a11y skill benchmarks across hosted and local model families. Runs cloud/Codex/Ollama benchmark scrip
Repro Auditor
Re-runs the code behind key thesis results in the worktree and compares numbers to the thesis. Use in /t2p-ass
Target Enricher
Researches the people behind the targets. Finds owners and management teams, tenure and succession signals fro
Related Skills
Is This A Scam
Paste it. Find out. Verdict in line one, the pattern in plain words, what to do, who to tell. Never shames, ne
Tab Rename
Rename the current terminal tab so you can tell your Claude Code tabs apart.
Find Unparked Items
Survey the workspace for RFDs and tasks that are actively being worked on — not shipped, not shelved, not aban