All authors
Claude Skills by bakw00ds
github.com/bakw00ds67 skills0 installs18 views
- Prompt Injection TestRun an OWASP LLM01 injection corpus against the system prompt + tool surface and report which payloads succeeded. Use when hardening an LLM agent's prompt/tools or auditing it for prompt-injection resistance.Votes: 0GitHub stars: 4
- Release AuditPre-release audit orchestrator — runs advisory multi-domain audits (Security, Code Quality, UI/UX, Docs, Performance, Regulated-Data, Mobile, Infra) across the project, produces versioned per-domain findings reports plus an executive summary with a lead-review prompt for human disposition. Two modes (audit / audit+remediate). Stack-aware — Phase 0 detects back-end / front-end / mobile profiles and dispatches only the relevant playbooks. Use whenever the user asks for a pre-release audit, rele...Votes: 0GitHub stars: 4
- Runbook AuthorGiven an alert definition (or a postmortem), produce a runbook entry — symptom, first 5 actions, escalation. Use when an alert lacks a response procedure or a postmortem action item calls for one.Votes: 0GitHub stars: 4
- Runtime PickGiven a task, recommend which yakOS agent + runtime to dispatch to (claude / codex / gemini). Use when unsure who should do a piece of work, before dispatching, to match task to specialist and runtime.Votes: 0GitHub stars: 4
- Sbom GenerateEmit a CycloneDX 1.6 or SPDX 3.0 SBOM for the project's locked dependency set, suitable for EO 14028 / enterprise procurement. Use when a customer, auditor, or compliance gate requires a software bill of materials.Votes: 0GitHub stars: 4
- Session RecoveryReconstruct working context after a session interrupt or fresh shell from on-disk artifacts. Use when resuming work and you've lost in-memory context — figure out where things stood.Votes: 0GitHub stars: 4
- Session SummaryEnd-of-session markdown report — agents dispatched, total cost, key decisions, time-to-completion. Use when wrapping a session, to produce a shareable record of what was done.Votes: 0GitHub stars: 4
- Source Driven DevelopmentGround framework- and library-specific implementation decisions in official documentation for the exact installed version, citing the source URL, instead of writing from memory. Use when implementing framework features (routing, forms, auth, state), writing reusable boilerplate, or any version-sensitive API call where an outdated or hallucinated pattern would be costly.Votes: 0GitHub stars: 4
- Split Mega TaskDecompose an over-large task into reviewable, independently-dispatchable chunks. Use when a single task turns out too big for one specialist or one reviewable PR.Votes: 0GitHub stars: 4
- Supervisor ToggleFlip the supervisor's block_on_critical setting (active ↔ passive) for the current project from the session. Use when the supervisor fires too aggressively (go passive) or the session enters higher-stakes work (go active).Votes: 0GitHub stars: 4
- Test Driven DevelopmentDrive implementation and bug fixes test-first (RED→GREEN→REFACTOR), proving behavior with a failing test before the code that satisfies it. Use when implementing new logic, changing existing behavior, fixing a reported bug, or adding edge-case handling — write the failing test first; a passing test is the evidence, "looks right" is not.Votes: 0GitHub stars: 4
- Test SuiteRun the full test suite with race detection and flake handling. Use when verifying a change broadly, before a PR or release, or when asked to run all the tests.Votes: 0GitHub stars: 4
- Tool Output CompactDeterministically shrink a large tool/command output (logs, JSON, search results) before an agent ingests it, via structural transforms — dedup, array-collapse, head+tail elision — not LLM summarization. Use when a command or tool produced a big output that would burn context, and you want a cheap, lossless-ish, repeatable compaction (use local-llm instead when you need semantic/meaning-level compression).Votes: 0GitHub stars: 4
- Usability ReviewHeuristic evaluation against a working prototype or live feature, rating issues on Nielsen's severity scale (cosmetic / minor / major / catastrophic). Use when assessing the usability of something already built and interactive.Votes: 0GitHub stars: 4
- Ux Writing ReviewVoice / tone / clarity audit of UI strings — action-oriented buttons, helpful errors, no jargon, consistent terminology, localization-friendly. Use when reviewing or polishing user-facing copy.Votes: 0GitHub stars: 4
- Verify Agent WorkSpot-check a teammate's claimed work before accepting it, against the actual artifacts. Use when a specialist returns a result and you need to confirm it's real before integrating or reporting it done.Votes: 0GitHub stars: 4
- Version BumpBump the project's VERSION file and prepend a CHANGELOG entry for the new version. Use when cutting a release or preparing one, to advance the version and record the changes.Votes: 0GitHub stars: 4