Diagnose Claude Code hooks, scripts triggered by events, when they fail to run or produce unexpected results. Follow checks for logs, registration, files, manual execution, hidden errors, and rebuilt TypeScript bundles to narrow down the cause.
Find unused functions through static analysis, which examines code without running it. Scan a project or directory, choose a language, and exclude designated entry points to get a list of functions with file locations.
Create a concise YAML handoff, a structured text document, for another agent or session. Capture completed work, decisions, blockers, file references, verification commands, and next steps so work can resume with the key context available.
Review saved session learnings to find recurring successes, failures, and decisions. Consolidates similar patterns, proposes reusable skills, rules, event-triggered actions, or agent updates, and creates approved changes with references to the source sessions.
Verify that infrastructure code is connected to the system and actually runs. Trace execution paths, check hook registration and database configuration, and test the full workflow to identify unused code or duplicate implementations before declaring completion.
Use this skill for end-to-end testing of a persistence pipeline, the sequence of steps that saves data. The supplied material states this purpose but provides only placeholder instructions, without concrete checks.
Review session changes, group related files into focused Git commits, and draft messages explaining their purpose. Presents the plan for approval before committing, omits Claude attribution, and records development reasoning through a project script.
Look up Claude Code commands and options for interactive work or automation. Covers non-interactive execution, session management, output formats, tool permissions, and agent definitions, with command examples for scripts and CI/CD workflows.
Inspect Claude Code execution traces in Braintrust, including tool calls and token usage. The guide explains event hooks, troubleshooting missing records, and matching separate parent and sub-agent traces using timestamps and metadata.
Review Claude Code sessions using Braintrust traces, records of agent activity. Summarize tool, agent, and skill usage, inspect repeated actions, replay sessions, and track token consumption to understand failures and recurring work patterns.
Coordinate background agents using system progress reminders while continuing other work. Read their output files after completion signals, avoiding repeated status checks and full transcripts that consume tokens and crowd the main conversation.
Is this your account? Link GitHub to bring it onto your creator page (Japanese)