Productivity●●●Banger
I mapped my AI coding setup – 90 of 103 installed skills never fire
Reveals that 90% of installed AI skills never fire, saving context window space and tokens.
Solve My ProblemDark Horse
Codefred
2014d ago
Back to browse
GitHub Repository
Evidence-based pruning for your AI-agent stack — scan, report, and safely prune unused skills/MCP servers/agents using real transcript evidence
47 starsGo
Skillreaper – find agent skills that load every session but never fire
by thousandflowers·Aug 4, 2026·1 point·0 comments
AI Analysis
●●●BangerSolve My ProblemDark Horse
Cuts 97% of dead token weight from agent context using transcript evidence.
Strengths
- •Evidence-based pruning uses real transcript data instead of guesswork.
- •Supports six platforms with zero telemetry and fully local execution.
- •Quantifies token waste per session to justify configuration changes.
Weaknesses
- •Only analyzes past transcripts; doesn't predict future tool needs.
- •Requires manual config file editing after receiving pruning report.
Category
Target Audience
Developers configuring AI agents with multiple tools and MCP servers
Similar To
MCP Inspector · LangSmith · Arize Phoenix
Similar Projects
Security●●●Banger
SkillGuard – scan agent skills for prompt injection payloads
First open-source scanner for AI agent skill supply-chain attacks.
Solve My ProblemDark Horse
arabking
213mo ago
AI/ML●●Solid
Agensi – Curated marketplace for AI agent skills (SKILL.md)
Security-scanned SKILL.md marketplace when GitHub repos have no vetting.
Niche GemSolve My Problem
Samuellrose
103mo ago
Developer Tools●●Solid
StackSkills – suggest agent skills from package.json
Scans package.json to recommend installable agent skills from the skills.sh ecosystem.
SlickNiche GemSolve My Problem
vab97
204mo ago
Security●●Solid
SkillsGuard – static scanner for malicious AI agent skills
Catches malicious AI skill scripts when Semgrep and Snyk miss the format.
Solve My ProblemShip It
teycir
201mo ago
Productivity●●Solid
Power Coding – an AI-coding skill I audited on Claude Code and Codex
Aerospace-style FMEA scans and handoff files keep AI agents from drifting off-course.
CozyBig Brain
anandp1989
2012d ago