Skill Claude CodeCodex
Use before major releases, after repeated cycles, or when relying on old ADRs, Gotchas, assumptions, QBR/M.AT calibration, or compound KB entries — automates staleness checks so evidence that was once true is revalidated, downgraded, retired, or escalated instead of silently trusted.
Skill Claude CodeCodex
Use when you have a written implementation plan to execute inline with review checkpoints — alternative to subagent-driven-development for environments without subagent support.
Skill Claude CodeCodex
Use before releasing, installing, or modifying Super Skills itself — self-audits the framework for broken skill references, YAML errors, stale version drift, nano mismatches, shell guard bugs, missing shared files, and DPS tooling availability.
Skill Claude CodeCodex
Use when searching for past decisions, pattern debt entries, risk assessments, or implementation plans — queries the super-skills KB using keyword search across ADRs, pattern-debt registry, and specs.
Skill Claude CodeCodex
Use after adr-commit completes — extracts lessons, domain terms, bug patterns, and gotchas from the current cycle into a structured compound wiki, making each cycle permanently improve future cycles.
Skill Claude CodeCodex
Use immediately after confirming any bug fix — before committing or moving to the next task. Prevents fixing one instance while leaving identical bugs elsewhere.
Skill Claude CodeCodex
Use before any task that touches PII, secrets, credentials, payment data, regulated data, logs, telemetry, external APIs, data export, multi-tenant boundaries, or C4 workflows — classifies data sensitivity, checks disclosure paths, scans for secret leakage, verifies redaction and retention, and blocks unsafe release…
Skill Claude CodeCodex
Use when receiving code review feedback, before implementing suggestions — requires technical verification not performative agreement, especially if feedback seems unclear or questionable.
Skill Claude CodeCodex
Use before merge, deploy, release branch cut, production traffic enablement, migration, or feature-flag rollout — verifies rollback, smoke tests, observability, data impact, dependency readiness, and post-deploy evidence so code-complete is not mistaken for release-ready.
Skill Claude CodeCodex
Use when a session must end before a task is complete, or when transferring work to a new agent session — captures current state compactly so the next session starts with full context rather than blank slate.
Skill Claude CodeCodex
Use once when starting a new project with super-skills — creates required directory structure, validates environment, and generates KB scaffolding. Run before any other super-skill.
Skill Claude CodeCodex
Use when completing tasks, implementing major features, or before merging — routes code review to the appropriate specialist lens (STRIDE, OWASP, ATAM, TEMPORAL, CPT, MIGRATION, PRIVACY/SECRETS) based on detected signals, with Independent Judge for Tier 3.
Skill Claude CodeCodex
Use when executing implementation plans with independent tasks — dispatches fresh subagent per task with specialist review between tasks, pattern-globalize after bugs, adr-commit at finish.
Skill Claude CodeCodex
Use when encountering any bug, test failure, or unexpected behavior — before proposing any fix. Requires root cause investigation with evidence anchors before attempting solution.
Skill Claude CodeCodex
Use after writing-plans produces a task list, during the Self-Review phase — scores each task for risk and ownership boundaries before handing off to execution.
Skill Claude CodeCodex
Use when implementing any feature, bugfix, migration, config change, or doc claim that needs proof — requires declaring a proof mode before implementation and preserves RED-GREEN-REFACTOR as the default for behavior code.
Skill Claude CodeCodex
Use when starting feature work that needs isolation from current workspace, or before executing implementation plans — ensures an isolated workspace exists, with harness-aware setup and adr-commit on finish.
Skill Claude CodeCodex
Use when starting any conversation or software-development task — establishes the Super Skills registry, runs complexity-gate for proportional rigor, and prevents rationalized skill skipping before acting.
Skill Claude CodeCodex
Use before claiming work is complete, fixed, passing, or done — requires running verification commands and producing fresh evidence before any success claim. Evidence before assertions, always.
Skill Claude CodeCodex
Use when you have an approved spec with audit-design PASS gate, before touching code — produces a comprehensive implementation plan with risk-scored tasks.
Skill Claude CodeCodex
Use when creating a new Super Skill, extending an existing one, or auditing skill quality — applies TDD-for-documentation with VHEATM quality lens.