All authors
Claude Skills by emgiezet
github.com/emgiezet29 skills0 installs0 views
- PlaybooksTask-scoped. Reached from the Task Router in `CLAUDE.md`. Do not read sibling playbooks. Load the `task-router` skill (one rule, one file) and `harness-eval` (how a trigger set is judged).Votes: 0GitHub stars: 3
- Agent OrchestrationHow to arrange agents for a task — the seven shapes (single, pipeline, fan-out, orchestrator/worker, handoff, debate, swarm), the control each one needs to stay safe, and the deterministic backbone that keeps the decision out of the model. Use when deciding whether to split work across agents, when a chain of agents loops or stalls, when a coordinator's context is the bottleneck, or when designing a system where an agent proposes and code decides.Votes: 0GitHub stars: 3
- Architecture ReviewVerify a planned architecture against the numbers it has to survive — the expected traffic profile, the availability tier (99.5 to 99.999) and what each tier forces in topology, the dependency-chain arithmetic, and the RPO/RTO the data path can actually deliver. Use when reviewing an arch.md or any architecture document, deciding whether a stated availability target is credible, writing a Non-functional targets block, or challenging a "five nines" request.Votes: 0GitHub stars: 3
- Code ReviewSeverity scale, verdict rule, review checklist, and spec-review lenses for pull-request inspection. Use when reviewing a PR, deciding a verdict, or writing a finding.Votes: 0GitHub stars: 3
- Dense TestingBuild the test grid that makes AI-written code safe — traceability from spec UC ids to tests, density floors per changed symbol, and a hard ban on skipping or weakening tests to get green. Use when deciding how many tests a change needs, wiring acceptance criteria to test names, generating tests for agent-written code, or when a suite is green but trust in it is low.Votes: 0GitHub stars: 3
- DiscoveryThe vocabulary and format rules for discovery artefacts — brief files, evidence tags, and the Definition of Ready. Use when writing or reading a brief, tagging a claim as evidence or assumption, deciding whether a ticket is ready to spec, or running any discovery command (/sdlc:brainstorm, /sdlc:discover, /sdlc:synthetic-users, /sdlc:backlog, /sdlc:ux-shape).Votes: 0GitHub stars: 3
- Fleet LoopRun the unattended issue-to-PR loop from inside an agent session — labelled issues in, reviewed draft pull requests out, every pipeline phase kept. Use when asked to "take issues from the tracker and ship PRs on their own", to keep `aisdlc loop` ticking under omp or Claude Code, or to explain why a tick pulled nothing. Not for one ticket with a spec (`aisdlc add`) or one issue by number (`aisdlc add --issue`).Votes: 0GitHub stars: 3
- Graphql Api DesignThe design rules for a GraphQL schema an agent can follow unattended — nullability as a contract, mutation input and payload shape, Relay connections, resolver cost and the N+1 trap, where a domain error belongs, and what counts as a breaking schema change. Use when adding or changing a type, field, query or mutation, reviewing a schema diff, writing a resolver, or judging whether a schema change is safe to ship.Votes: 0GitHub stars: 3
- Harness EvalMeasure and tune the AI SDLC harness so the pipeline delivers on a cheap model — run scenario × model evals against the sandbox fixture, then fix the failing instruction rather than raising the model tier. Use when a task only works on the expensive model, when adding or changing a playbook or rule, when per-PR cost is climbing, or when writing a new eval scenario.Votes: 0GitHub stars: 3
- Pipeline ContractsThe shared vocabulary every /sdlc command after ship and before spec speaks — tracker and browser descriptors, the operation names, chain markers, verdict tokens, the claim lock, the pipeline labels, and the bugfix approval rule. Use when a command needs to talk to the tracker or a browser, hand a PR or issue number to the next command, decide whether it may take a PR another agent claimed, set or read a pipeline label, or decide whether a bugfix spec counts as approved.Votes: 0GitHub stars: 3
- Rest Api DesignThe design rules for an HTTP/REST surface an agent can follow unattended — resource naming, status-code selection, one error envelope per API, bounded pagination, idempotency, and what counts as a breaking change. Use when adding or changing an HTTP endpoint, reviewing an OpenAPI or contract diff, deciding a status code or a URL shape, or judging whether an API change is safe to ship.Votes: 0GitHub stars: 3
- Security ReviewThe lens a reviewer applies to a diff to find security findings — trust boundaries, the ten review lenses, what counts as a finding and what does not. Use when reviewing a PR that touches authentication, authorization, session handling, secrets, input validation, or external requests.Votes: 0GitHub stars: 3
- Spec AuthoringWrite specs that are fuel for agents rather than documents for humans — a fixed section layout with stable, testable UC ids that drive implementation, tests, and QA. Use when writing or reviewing a spec in specs/<TICKET>/spec.md, turning a ticket or a written request into acceptance criteria an agent can execute unattended, deciding whether a spec is ready to queue, or splitting an oversized spec.Votes: 0GitHub stars: 3
- Task RouterMaintain a project's instruction hierarchy — a small CLAUDE.md routing table plus task-scoped playbooks and path-scoped rules — so agents load only what the task needs. Use when a CLAUDE.md has grown too large, agents ignore or contradict instructions, token cost per task is rising, or when adding a playbook, rule file, or router row to a project.Votes: 0GitHub stars: 3
- Agent DisciplineAlways-on agent behavior rules: linter suppression policy, required test gates, test deletion, remote scripts and look-alike packages, credential handling, scope discipline.Votes: 0GitHub stars: 3
- Auth AntipatternsForbidden authentication and session-management patterns — weak token generation, insecure password hashing, session fixation, JWT algorithm confusion, missing rate limits, and account enumeration. Applies when writing auth, login, session, token, password, or OAuth code.Votes: 0GitHub stars: 3
- Csharp AntipatternsForbidden C# patterns — SQL injection, BinaryFormatter, weak random, async void, exception swallowing. Applies when editing C# files.Votes: 0GitHub stars: 3
- DeslopRefactors a file or directory to remove AI slop — duplication, dead code, speculative abstraction, bloated functions — without changing observable behaviour. Takes a path argument. Invoke with /slop-guard:deslop <path>.Votes: 0GitHub stars: 3
- Go AntipatternsForbidden Go patterns — SQL injection, shell injection, missing timeouts, resource leaks, goroutine leaks, and error handling issues. Applies when editing Go files.Votes: 0GitHub stars: 3
- Iac AntipatternsForbidden IaC/container/CI patterns — public cloud resources, unencrypted storage, privileged containers, unpinned images, insecure CI workflows.Votes: 0GitHub stars: 3
- Jvm AntipatternsForbidden JVM (Java/Kotlin) patterns — JDBC injection, insecure deserialization, weak random, trust-all TLS, Jackson misconfiguration.Votes: 0GitHub stars: 3
- Node AntipatternsForbidden Node.js backend patterns — prototype pollution, shell injection, path traversal, ReDoS, sync I/O in handlers. Applies when editing JS/TS Node files.Votes: 0GitHub stars: 3
- Php AntipatternsForbidden PHP/Laravel/Symfony patterns — SQL injection, shell injection, weak crypto, N+1, and maintainability anti-patterns. Applies when editing PHP or Blade files.Votes: 0GitHub stars: 3
- Python AntipatternsForbidden Python patterns — SQL injection, shell injection, unsafe deserialization, missing timeouts, blocking async calls. Applies when editing Python files.Votes: 0GitHub stars: 3
- Ruby AntipatternsForbidden Ruby/Rails patterns — eval with user input, YAML.load, mass assignment without permit, shell injection. Applies when editing Ruby files.Votes: 0GitHub stars: 3
- Rust AntipatternsForbidden Rust patterns — unsafe without safety comment, unwrap in library code, unchecked integer arithmetic. Applies when editing Rust files.Votes: 0GitHub stars: 3
- Secure ReviewSecurity review subagent: reads session findings and the AP-* catalogue, produces a structured security assessment. Cannot write or edit files — a reviewer that can edit is not a reviewer. Invoke with /secure-review.Votes: 0GitHub stars: 3
- Sql AntipatternsForbidden SQL patterns — SELECT *, missing CONCURRENTLY on indexes, unsafe migrations, missing FK indexes. Applies when editing SQL files.Votes: 0GitHub stars: 3
- Ts React AntipatternsForbidden TypeScript/React patterns — XSS, eval, localStorage tokens, hook bugs, floating Promises. Applies when editing TS/TSX files.Votes: 0GitHub stars: 3