Oleks
06cb5b86ff
commands/offload-audit.md tells a heavy-target caller to hand the work to Agent(offload-analyst). Nothing told the analyst it IS that delegation target, so on a large plugin it re-judged the target heavy by the same criteria and spawned another offload-analyst instance, which did the same again — observed live against the cluster plugin (11 agents, ~27 skills), 3-4 levels deep, each restarting the same early measurement steps with no forward progress. Fixed by making the analyst explicit that its only Agent calls are the fixed filing chain, and scoping the command's delegation guidance to the top level only.
1.7 KiB
1.7 KiB
description, argument-hint, allowed-tools
| description | argument-hint | allowed-tools | ||||||
|---|---|---|---|---|---|---|---|---|
| Audit a Claude Code plugin for steps that should be a deterministic script instead of raw inference — measure, classify, file the well-evidenced candidates, and write the run summary. Usage — /offload-audit <plugin|path> [--days N] | <plugin name or path> [--days N] |
|
/offload-audit
Run the full audit on $ARGUMENTS.
- Parse
$ARGUMENTSinto a target (a plugin name or a directory path) and an optional--days Nwindow, defaulting to 30. - If no target was given, ask which plugin to audit. Do not guess one, and do not default to the current directory — auditing the wrong plugin wastes a full transcript scan.
- Invoke the
offload-auditskill and follow it exactly, including both filing gates (the overrule case, FR-4.2; the evidence-stability check against the previous window, FR-4.5) and the marker search. For a heavy target, or when the report should not consume this session's context, hand the work toAgent(subagent_type="inference-arbitrage:offload-analyst")instead — one agent for the whole audit, never one per skill (spec §7). This delegation happens at most once, from this top-level command to the analyst. The analyst instance you spawn is the delegation target, not another decision point — it never re-delegates the whole audit to a furtheroffload-analystinstance, however large the target looks from inside.
Filing is implicit: a candidate that clears the rubric gate and the value threshold is filed without asking. Everything below the gate is reported as a boundary question and filed nowhere.