Primary editorial path for this compatibility group. The full instructions and support files remain local so existing installations
continue to work offline. This is one shared procedure, not an additional capability.
Preserve the callable ID when an existing manifest or client configuration uses it.
Modified in AAS on 2026-09-05; original metadata and license notices are retained.
Review a defined diff or subsystem from several relevant perspectives, such as
correctness, authorization and performance. The skill describes a review procedure;
it does not install an orchestration engine or prove compliance.
Pin repository/base/head, the changed paths, intended behavior and available tests.
Use independent agents only if the user authorizes delegation and the host supports
it. Otherwise conduct the perspectives sequentially. Do not spawn agents merely
because this callable ID contains “multi-agent”. Read-only review is the default;
fixes, external posts and deployment stay within the user's actual task authority.
bounded question, owned paths, time/effort limit and expected evidence format.
hypotheses separate from demonstrated failures; do not invent confidence scores.
repeating a claim are not evidence of correctness.
the original failing result; a rerun does not erase it.
For a tenant-scoped cache change, review key construction, authorization context and
invalidation. Reproduce two tenants requesting the same prompt and different prompt
versions within one tenant. Expected result: a finding only if an actual cross-scope
hit or stale result can occur, with the smallest case showing it. A general style
opinion must not be presented as a data-exposure defect.
No agent-routing code, compliance validator or quality-score calculator is bundled.
Parallel review adds cost and can duplicate blind spots. This procedure cannot prove
whole-repository safety from a diff, nor authorize production load tests or messages.