whereismyai
Rashomon
Rashomon is an audit tool for Claude Code that keeps an independent record of agent actions to compare against the agent's summary.
Rashomon tracks and independently records every action taken by Claude Code, including executed commands, file edits, and their success or failure status. It contrasts this record directly against the agent's closing summary, alerting the user if a step failed and the closing message omitted failure-related words. The tool also exposes helper subagent actions that normally stay hidden from the primary transcript. It operates locally on macOS and Linux without sending prompts, code, or outputs over a network. Designed specifically for Claude Code users, it functions quietly in the background and only alerts developers when discrepancies or failures occur.
Rashomon at a glance
Key features
- Recording tool calls, commands run, and file edits
- Tracking hidden subagent actions and executions
- Detecting test runner results like go test, pytest, and npm test
- Comparing agent summaries against a list of 43 failure words
- Reporting command rewrites and visibility limitations
- Running offline with zero telemetry or network calls
- Operating as a Claude Code plugin or via command line
Use it for
- Catching failed test runs hidden by inaccurate agent summaries
- Auditing the hidden background activities of helper subagents
- Verifying Claude Code tool execution accuracy on macOS or Linux
Companions
Tools that articles about Rashomon recommend using with it, each with its job.
- Claude Code — Built to work with Claude Code.
- Context7 — Provides up-to-date documentation for AI agents and code editors.
- Playwright MCP — Enables LLMs to interact with web pages using structured accessibility snapshots.
- GitHub MCP Server — Reads pull requests, issues and code search on GitHub.
- Superpowers — Brainstorm, plan and execute skills for bigger tasks.
- gstack — Skill bundle for planning, review and shipping.
Alternatives to Rashomon
- Reviewer Agent — It uses a separate AI model to review code and prevent premature agent stopping.
- Flash Agents — Offloads reading and typing to cheap workers while keeping Claude for review
- Agent Archive — Hand off coding sessions between Claude Code, Codex, and Cursor across machines.
- Runway — Control Claude Code and Codex sessions on local or cloud machines from any phone.
- Claude Super Sidebar — It brings agent relationships, execution states, and task flows into one panel.
- Tracon — Spots stuck subagents and worktree file conflicts before your merge.
- Habblaud — Displays Claude Code agent activities and resource usage in a virtual office.
- Skyborne — Track multiple Claude Code agent sessions and handle approvals in one local 3D interface.
- Ranwhat — Audits coding agent actions and scans for secrets on your machine without telemetry.
- claude-mem — It shares persistent, searchable agent memory across sessions and devices.
- Oh My ClaudeCode — Orchestrates parallel Claude, Gemini, and Codex CLI workers.
- Stroq — Scans recorded sessions and guards future actions with local hooks.
All Rashomon alternatives