Skillreaper is an open-source skill lifecycle management and auditing tool for AI agent ecosystems, developed by thousandflowers. It analyzes agent session transcripts and execution histories to identify, evaluate, and prune unused, broken, or obsolete agent skills and plugins. Operating as a Go-based CLI utility (`reap`) and packaged with Claude plugin support, Skillreaper evaluates tool invocations across retention windows. It categorizes tools into explicit verdicts—such as REAP(unused), REAP(broken), REVIEW(rare), and REVIEW(needs-data)—by comparing recent session activity against durable execution history to prevent accidental pruning of infrequent but essential tools. Unlike execution-focused agent harnesses that only manage tool runtime dispatch, Skillreaper serves as a maintenance harness to keep agent context windows lean and toolsets healthy. It prevents context bloat and tool selection degradation by providing systematic hygiene over expanding agent capabilities. Skillreaper is open-source software distributed via GitHub under standard open-source terms.
Transcript Session Auditing — Scans multi-session interaction logs and execution traces to evaluate tool firing frequency across configurable retention windows. - Heuristic Verdict Generation — Outputs granular status classifications including REAP(unused), REAP(broken), REVIEW(rare), and REVIEW(needs-data). - Durable History Consultation — Cross-references long-term records to differentiate between truly abandoned tools and low-cadence utilities. - Error-State Isolation — Flags consistently failing skills as broken regardless of historical success, surfacing defective tools for remediation. - Claude Plugin Support — Includes built-in .claude-plugin configuration metadata for native integration in Claude agent environments. - Standalone Go CLI — Provides a lightweight reap binary suitable for local execution and automated CI/CD maintenance workflows.
Use cases
Use Case: Agent Context Optimization Audit extensive plugin and skill catalogs across agent workspaces to safely remove unused capabilities, reducing tool-definition token overhead in agent prompts.
Use Case: Automated Broken Tool Detection Identify plugins and tools that consistently error out in agent transcripts, generating clear failure verdicts for maintainers to fix or deprecate.
Use Case: Safe Skill Retention Auditing Evaluate low-cadence utility skills across multi-week evaluation windows, ensuring that rarely used but critical recovery tools are preserved rather than pruned as abandoned.
Who it is for
Designed for developers, agent framework maintainers, and platform teams operating large skill registries or modular Claude agent setups. It is not meant for single-prompt workflows or simple agents with small, static tool definitions.
Install guide
Installation Install the reap binary using Go: bash go install github.com/thousandflowers/skillreaper/cmd/reap@latest Alternatively, clone and build the repository locally: bash git clone https://github.com/thousandflowers/skillreaper.git cd skillreaper go build -o reap ./cmd/reap
First Run Verify that the binary is available in your $PATH and check the available commands: bash reap --help
Example Session Run an audit against your agent session history to inspect skill health and obtain pruning recommendations: bash # Audit skill invocations against the standard session transcript window reap
hol-guard — hol-guard is a security scanner and agent guardrail harness developed by Hashgraph Online. Built to safeguard agentic runtime environments and…
Google AX — Google AX is an open-source agent harness and execution orchestrator developed by Google. It provides runtime infrastructure and execution sandboxing…
Agent Island — Agent Island is an open-source agent orchestration harness created by developer tristan666666 on GitHub. Designed as a framework for running…
ActPlane — ActPlane is an open-source agent execution and policy enforcement harness developed by the Eunomia-bpf organization. It functions as a security…
Stratless — Stratless is an open-source agent harness and execution orchestrator developed by the Stratless AI team. It provides a modular infrastructure…
NemoClaw — NemoClaw is an open-source agent orchestration harness and execution framework developed by NVIDIA. It provides structured runtimes, modular skill dispatching,…