alinafe82

agent-security-boundary

Review agent workflows when trust boundaries, data access, tool permissions, or external side effects change.

alternatives-before-code

Compare viable approaches when an unresolved design decision has meaningful cost or reversibility tradeoffs.

assumption-audit

Audit consequential unverified assumptions in a plan or proposed explanation.

complexity-budget

Review whether proposed abstractions or infrastructure justify their maintenance and operational cost.

debugging-lab-notebook

Investigate hard or intermittent failures by tracking hypotheses, experiments, and regression evidence.

diff-interrogation

Review a code diff for behavioral regressions, missing evidence, and security or data risks.

docs-claim-audit

Verify changed public documentation claims against implementation and recorded checks.

evidence-to-test

Turn a specific recurring review failure or validator gap into a repeatable regression check.

explain-without-ai

Coach a developer through explaining and testing the mechanism behind a change.

failing-test-first

Reproduce a reported bug, fix it, and verify the regression with the narrowest useful test.

god-mode

Coordinate relevant workflows when the user explicitly requests god mode or multi-skill orchestration.

problem-framing

Resolve an ambiguous engineering request into an observable problem and completion criteria.

read-the-docs-first

Verify version-sensitive API behavior or a documented repository constraint needed for the current task.

release-readiness

Check versioning, distribution metadata, licenses, and validation evidence for a planned skills release.

runtime-adapter-smoke

Validate skill discovery and loading after runtime adapter or manifest changes.

skill-authoring-gate

Review a new or materially changed skill for distinct purpose, useful routing, and executable guidance.

skill-deprecation-review

Assess whether an existing skill should be retained, merged, renamed, or retired, including migration impacts.

skill-overlap-audit

Compare overlapping skill triggers and workflows to resolve ambiguous selection.

thinking-ledger-review

Review an existing thinking ledger against its cited evidence and applicable risk requirements.

trace-the-code

Trace an unfamiliar execution path when a change depends on its state, side effects, or error handling.

transcript-review

Audit a real agent transcript for unsupported claims, missed evidence, and instruction failures.

good-skill

Validate a well-formed fixture skill. Use when testing the validator happy path. NOT for production assistant behavior.