By edo-ceder
Agent-and-human product lifecycle for app.st4ck.io workspaces, with role-specific workflows, deterministic QA, and server-enforced test lineage and attestation.
Debug and fix bugs with role-separated agents — research, test gaps, fix, verify green. You are the dev manager.
Phase 5 §5.2 — agent-driven test impact analysis. Reads the local git diff, surfaces every QA test whose components cite the changed lines, queues test_design_change dev_tasks for QA. Optionally generates proposed component updates with --propose.
Full feature lifecycle — requirements → plan → code → QA → deliver. Role-separated agents with human gates.
Migrate legacy tests to v2 component format. Activates the qa-testing-migration skill, which classifies each test by shape (agentic / components_v1 / components_v2 / mixed) via classify_test_migration_shape and runs the appropriate internal branch (agentic re-author OR component-upgrade) inline. Per plan §13.2 (consolidated 2026-04-26 — was three skills before).
PO feature research — explore codebase, challenge requirements, surface logic gaps, present solution options with effort/risk.
Use this agent to implement features from the plan's task list. Writes code, unit tests for pure functions, and runs quality gates. Never writes or reviews QA tests.
Use this agent to review code changes for bugs, logic errors, security vulnerabilities, and adherence to project conventions. Returns findings with confidence scores.
Use this agent to explore and understand a codebase — traces architecture, patterns, dependencies, similar features, data models, and git history. Returns structured findings for planning.
Independent developer reviewer for a Product Requirements Document. Verifies source provenance, spot-checks claims against the cited source code/export, and identifies dead-code claims or shallow decomposition. Used as one of three reviewer angles by the prd-review skill. Cannot modify the PRD — read-only with structured report output.
Independent PO/stakeholder reviewer for a Product Requirements Document. Reads the PRD from a non-technical product-owner perspective and reports findings with severity and source citations. Used as one of three reviewer angles by the prd-review skill. Cannot modify the PRD — read-only with structured report output.
Forwarding stub. This skill was consolidated into qa-testing-migration on 2026-04-26 (plan §13.2, mid-12). Activate qa-testing-migration instead — it covers Path A (heavy re-author) AND Path B (light component-upgrade) inline, with classify_test_migration_shape selecting the branch per test.
Phase 5 §5.2 — agent-driven test impact analysis. Reads the local git diff, calls get_tests_affected_by_diff, creates test_design_change dev_tasks per affected component. Optionally invokes the propose subworkflow to attach proposed component diffs. Triggered explicitly via `/st4ck:impact` or auto on PostToolUse(Edit/Write) hits inside a cited file.
Forwarding stub. This skill was consolidated into qa-testing-migration on 2026-04-26 (plan §13.2, mid-12). Activate qa-testing-migration instead — it covers Path A (heavy re-author) AND Path B (light component-upgrade) inline, with classify_test_migration_shape selecting the branch per test.
Create comprehensive implementation plans for features including phased tasks, security analysis, test strategy, and migration files. Use when transitioning from approved requirements to implementation.
Author a Product Requirements Document from an existing codebase, code export, or unpacked low-code project — by reading source rather than asking the user. Use when the user wants a PRD generated from imported elements, exported builders (Bubble, Webflow, Retool), screen recordings, or any "code-first" reverse-engineering of intent. Self-contained — bundles its foundational PRD-authoring rules so it works with or without a server connection.
Uses power tools
Uses Bash, Write, or Edit tools
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.
The agent-and-human lifecycle surface for full st4ck — requirements through code, QA, and delivery.
st4ck has two product surfaces: st4ck-lite, the local OSS surface, and full st4ck, which pairs this plugin with an app.st4ck.io workspace. Full st4ck builds on Lite's PRD authoring, agent-driven recording, and deterministic replay and adds agent-team lifecycle orchestration, persistent intent, test/component lineage, attestation, impact analysis, and shared reuse. The MCP-backed lifecycle features require a workspace; the plugin is their agent-and-human client, not a standalone middle tier.
# Inside Claude Code:
/st4ck:implement "Add a per-team usage cap to the billing portal" # full feature lifecycle
/st4ck:debug "Profile edits aren't appearing in the activity feed" # role-separated debug flow
/st4ck:po-research "Should refunds be reversible after settlement?" # PO discovery + options
/st4ck:impact # diff → which tests are affected
Lifecycle commands coordinate purpose-built subagents, explicit human checkpoints, and role-specific write restrictions where configured. Those controls reduce accidental role overlap; the authoritative lineage, attestation, execution, and required-review gates are enforced by the st4ck service.
| Skill | Purpose |
|---|---|
/st4ck:po-research | Product-owner feature research — explores codebase, challenges requirements, surfaces logic gaps, presents solution options with effort/risk |
/st4ck:plan-author | Comprehensive implementation plans — phased tasks, security analysis, test strategy, migration files |
/st4ck:implement | Full feature lifecycle (requirements → plan → code → QA → deliver) with role-separated agents + human gates |
/st4ck:debug | Role-separated debug flow — research, test gaps, fix, verify green. The session is the dev manager |
/st4ck:impact | Agent-driven test impact analysis: reads local git diff, surfaces affected QA tests, queues test_design_change dev_tasks |
/st4ck:supervise | Pause + regroup — reviews the session transcript, reminds you what you said, what's left |
/st4ck:regression-author | Author regression test suites for shipped features |
/st4ck:regression-run | Execute signed regression suites and report results with evidence (observe-only) |
/st4ck:version-author | Author version tests for in-development features (green as implementation lands) |
/st4ck:qa-testing-debug | Diagnose + fix a failed test or component |
/st4ck:st4ck-run | Execute a deterministic test using the runner (with agentic-block IPC pause + rerun from failure) |
/st4ck:st4ck-browse | Drive a real browser one IPC primitive at a time, optional --record saves as a deterministic md test |
Source-grounded Product Requirements Document authoring — useful when a platform has been live for a while and the test/spec layer needs an intent anchor.
| Skill | Purpose |
|---|---|
/st4ck:prd-from-source | Author a PRD by reading source rather than asking the user. Four iron rules; Phase 0 preliminaries (checks get_project_users, asks user for role anchor + extra docs); two-pass mechanical-scaffold + curated-intent approach; cross-module reuse rule; audience triple-target (non-technical / QA / dev); confidence markers (CONFIRMED / INFERRED / GAP). |
/st4ck:prd-review | Four-phase review pipeline — self-review → 3 parallel independent reviewers (PO / QA / Dev) → known-gaps-vs-code pass → bug-routing to dev tasks. Converges on diminishing returns. |
Three reviewer subagents — prd-reviewer-po, prd-reviewer-qa, prd-reviewer-dev — encode each review angle as a stable agent type for consistent, cached prompts across sessions.
code-agent, code-reviewer, codebase-explorer, qa-author, qa-reviewer, qa-runner, solution-analyst — each with a defined workflow role and role-specific tool/write settings. These client-side settings are workflow guardrails, not a security boundary.
1. Role separation by workflow and configured write scopes. No agent is asked to do everything. The qa-author is configured without direct file-edit tools, the code-agent is not given st4ck QA tools, and the code-reviewer is configured without Edit/Write. Prompts and client tool settings reduce accidental overlap, but they are not a sandbox or security boundary.
npx claudepluginhub edo-ceder/st4ck-plugin --plugin st4ckRecord E2E tests with your AI agent. Replay them forever, for free. No account, no key, no service — just md files in your repo.
A growing collection of Claude-compatible academic workflow bundles. Covers scientific figures, manuscript writing and polishing, reviewer assessment, citation retrieval, data availability, paper reading, literature search, response letters, paper-to-PPTX conversion, and evidence-grounded Chinese invention patent drafting. Rules are organized as reusable skill folders with explicit workflows and quality checks.
Consult multiple AI coding agents (Gemini, OpenAI, Grok, Perplexity, plus codex, antigravity, and grok CLIs when installed) to get diverse perspectives on coding problems
Comprehensive 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.
Tools to maintain and improve CLAUDE.md files - audit quality, capture session learnings, and keep project memory current.
Comprehensive PR review agents specializing in comments, tests, error handling, type design, code quality, and code simplification
Upstash Context7 MCP server for up-to-date documentation lookup. Pull version-specific documentation and code examples directly from source repositories into your LLM context.