By vancourse
Opinionated SDLC framework — plan before coding, reconnaissance before design, adversarial design checklist, phase-routed skill gates, hooks for workflow automation. Pairs custom review skills with curated third-party plugin dependencies.
[USER] Dispatch a review skill as two parallel worktree-isolated agents (cooperative steelman + adversarial attacker) and reconcile their findings — the canonical two-voice review pass
[USER] 7-pass foundational architecture review — framework, persistence + tenancy, trust boundaries + auth + data classification, observability + deploy, design system, branch/release, system-level threat model. Each pass produces an ADR. Plus adversarial counter-pass.
[USER] PR Copilot review loop — GraphQL trigger (REST + gh CLI silently no-op on bots), GraphQL verify, address-and-rerequest cycle, stop at NITs-only OR 5 rounds
[USER] Feature-design-from-scratch gate — scenarios, MVP cut, deferral list, integration, failure modes; plus product-overlap + OSS-alternatives + adversarial counter-pass
[USER] Profile this project's bug-loop tooling (issue tracker / observability / reproduce) into a routing-only, secret-free apex.profile.toml — probes deps, CI configs, connected MCP servers, and the git remote; auto-fills what it can infer, asks for the rest
Review a single Architecture Decision Record (ADR) before locking it in or accepting an amendment. 5-element audit — context, decision, alternatives, consequences (including security + cost + reversibility), status field. Plus adversarial counter-pass. Pairs with apex:architecture-design (produces 7 ADRs at project start) and runs at every amendment. Fires when authoring or reviewing an ADR — initial set, amendment to an existing one, or supersession. Keywords: ADR, architecture decision record, amendment, supersession, decision log.
Canonical dispatch mechanic for running any apex review skill (design, plan, implementation, PR) as two parallel worktree-isolated agents with opposite framings — cooperative steelman + adversarial attacker — then reconciling their findings. Promoted from prose in CLAUDE.md / design-feature into a first-class skill so every phase-routed review skill points to one source of truth for HOW to dispatch the pair. Pairs with apex:design-review, apex:impl-plan-review, apex:ai-pre-review-checklist, apex:threat-model (any of which can be the review skill the pair runs). Fires after design, plan, or implementation of a non-trivial change, before declaring the artifact frozen / ready / done. Keywords: adversarial pair, parallel review, cooperative agent, adversarial agent, steelman, attack pass, two-agent review, worktree review, cold pass.
9-step pre-PR robustness gate for any AI-assisted branch — force branch explanation, layering, state ownership, concurrency, success/failure/fallback, test quality, consumer-tracing, reviewer simulation, unvalidated gaps. Run this BEFORE you reach `git push`. The skill walks each step's full prompt against an assistant. Fires when a feature branch is ready for review and AI wrote or refactored code. Keywords: pre-review, ai-assisted, pr readiness, reviewer simulation, robustness gate, before push.
Opinionated SDLC framework — plan before coding, reconnaissance before design, adversarial design checklist with alternatives and critiques, phase-routed skill gates for planning/implementing/reviewing, port-verification discipline. Fires when planning a non-trivial change, designing a new endpoint or service, refactoring across files, shrinking a bloated PR or design, adding support for a new scope/source/kind/variant, opening a PR, or reviewing changes. Keywords: plan, design, refactor, endpoint, payload, review, PR, implement, ship, shrink, bloated, minimal diff, support a new, subtractive design, reconnaissance, recon.
Reviews API endpoints and response payloads from the *consumer's* perspective — does each field earn its place, is the error shape actionable, is the function over-scoped, are timeouts justified. Complements python-review / typescript-review (which check producer-side correctness) by adding the "does this surface deserve to exist as designed" lens. Fires when reviewing a new endpoint, a new request/response Pydantic/Zod model, a new service handler, or a payload definition. Keywords: api review, endpoint review, payload shape, response model, request model, api design, rest, handler, service function.
Executes bash commands
Hook triggers when Bash tool is used
Modifies files
Hook triggers on file write and edit operations
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
An opinionated Claude Code plugin that bundles a personal SDLC framework — planning gates, code-review skills, PR workflow discipline, and workflow-automation hooks. Designed to make AI-assisted coding survive a strict reviewer.
New to apex? Start with HOWTO.md for install + the 6-command quickstart, then WALKTHROUGH.md for the idea → shipped product/feature path in order. See FLOW.md for the phase-by-phase flowchart and skill × phase matrix.
Already using a spec-driven tool? apex composes with GitHub Spec Kit and BMAD-METHOD — author specs with them, then run apex's adversarial review/freeze gates on their artifacts. See INTEROP.md.
Contributing to apex itself? See MAINTAINING.md for maintainer-only discipline (pair-pattern verification, slash-menu-count grep, etc.). The PreToolUse hook surfaces it automatically when you edit skills/, commands/, hooks/, or rules/ inside the apex plugin repo.
For when each skill fires, see FLOW.md. This table is what each skill does.
| Skill | What it does |
|---|---|
architecture-design | 7-pass foundational architecture review (framework / persistence + tenancy / trust boundaries + auth + data classification / observability + deploy / design system / branch + release / system-level STRIDE threat model). Each pass outputs an ADR. Architecture FREEZES after all 7 ADRs are accepted. One-time at project start; re-runs only on amendment-triggering changes. |
adr-review | Review a single Architecture Decision Record. 5-element audit — context, decision, alternatives (≥2 real), consequences (incl. security + reversibility), status field. Fires for every ADR (initial set + amendments). |
prd-review | 7-pass PRD audit (acceptance criteria / testable scenarios enumerated — each tagged by highest test layer (integration / E2E); compound ones optionally broken into use-case one-liners / out-of-scope / unknowns / success metric / sequencing / spec-freeze readiness) + internal-product-overlap scan + OSS-alternatives scan + inline adversarial counter-pass. The SPEC-phase gate, before design. |
apex-flow | Umbrella planning gates: §1a reconnaissance + §1b adversarial design checklist + §1c verify ask vs raw quotes + phase-routing pointer |
recon | Reconnaissance brief before design — promotes apex-flow §1a from an in-head checklist into a first-class artifact. Scoped to the change's blast radius: enumerate the authoritative primitives that already answer the design's questions, distill their contracts (not signatures), capture invariants + trust boundaries, run the producer/consumer + first-affordance checks against that fact base, and persist durable semantic facts to domain-knowledge. Emits a short Recon Brief that design-feature / §1b consume. Auto-nudged by suggest-skill-on-prompt on subtractive-design traps ("shrink/bloated X", "support a new scope/kind/variant"). |
design-feature | Feature-design-from-scratch gate (NEW features, not fixes) — scenarios + MVP cut + deferral list + integration with existing surface + failure modes + §6 attack surface (invokes apex:threat-model), with product-overlap + OSS-alternatives + adversarial counter-pass. Distinct from apex-flow §1b which is generic. |
threat-model | Per-feature STRIDE threat model at design time — Spoofing / Tampering / Repudiation / Information disclosure / Denial of service / Elevation of privilege, against the feature's attack surface anchored on architecture-design Passes 3 + 7. Outputs a "Threat Model" section appended to the design doc that security-review audits at PR time. |
observability-review | Per-feature design-phase observability-contract gate (5 passes) — structured logging (one log per failure decision, correlation-tagged) / metrics + cardinality budget (each feeds a named SLI) / tracing + causality across sync + async boundaries / alerting + SLO (page on symptoms not causes, runbook + owner) / privacy in telemetry (no PII/secrets in logs, metrics, traces). Asks "when this misbehaves in prod, can an operator SEE why?" Instantiates the stack + SLO/alerting policy from architecture-design Pass 4 per-feature; distinct from security-review Pass 5 (security-event slice at PR time). Outputs an "Observability" section the design doc carries. Plus adversarial counter-pass. |
npx claudepluginhub vancourse/apex --plugin apexComprehensive skill pack with 66 specialized skills for full-stack developers: 12 language experts (Python, TypeScript, Go, Rust, C++, Swift, Kotlin, C#, PHP, Java, SQL, JavaScript), 10 backend frameworks, 6 frontend/mobile, plus infrastructure, DevOps, security, and testing. Features progressive disclosure architecture for 50% faster loading.
Harness-native ECC operator layer - 67 agents, 271 skills, 92 legacy command shims, reusable hooks, rules, selective install profiles, and production-ready workflows for Claude Code, Codex, OpenCode, Cursor, and related agent harnesses
Develop, test, build, and deploy Godot 4.x games with Claude Code. Includes GdUnit4 testing, web/desktop exports, CI/CD pipelines, and deployment to Vercel/GitHub Pages/itch.io.
Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques
Tools to maintain and improve CLAUDE.md files - audit quality, capture session learnings, and keep project memory current.
Access thousands of AI prompts and skills directly in your AI coding assistant. Search prompts, discover skills, save your own, and improve prompts with AI.