agent-delegation-audit

Reviews a repository of Claude Code plugins, skills, and commands to find work that belongs in sub-agents — operations whose output floods the conversation, tasks that deserve restricted tool access, and jobs misaligned with model tier. Compares the findings against existing agent definitions to surface missing agents, over-broad tool grants, duplicate agents, off-tier assignments, and stale delegation references, then delivers a prioritized report and, once the user approves specific changes, writes the agent files and syncs the agent README and the delegation mapping table. Use when asked to assess sub-agent coverage, cut transcript noise from chatty commands, right-size agent models and tool sets, or repair delegation mappings. Not for executing the audited builds, tests, or scanners, for editing agent definitions without explicit approval, or for tuning one skill's internal prompts.

Version

v1.0.0

Size

15 KB

Updated

2026-10-10

Your client verifies the signature before reading. This explicit manual read is available on every plan. Free automatically loads five chosen skills; paid plans automatically load relevant skills across the library.