{
  "markdown": "# ProdMaster AI\n\n![Version](https://img.shields.io/badge/version-2.1.0-blue) ![License](https://img.shields.io/badge/license-MIT-green) ![Platform](https://img.shields.io/badge/platform-Claude%20Code-orange) ![Skills](https://img.shields.io/badge/skills-31-purple)\n\n> The engineering ops layer for Claude Code — one command orchestrates features, QA, code review, deployment, security audits, and performance benchmarks, while the plugin self-improves based on your velocity data.\n\n**Install:**\n```bash\nclaude plugin install prodmaster-ai@prodmaster-ai-marketplace\n```\n\n**Or via the official marketplace (once listed):**\n```bash\nclaude plugin install prodmaster-ai@claude-plugins-official\n```\n\n---\n\n## Why ProdMaster AI\n\nMost Claude Code workflows require juggling many commands across many contexts. ProdMaster AI collapses that into one entry point:\n\n- **`/prodmasterai build auth`** — breaks the feature into tracked parallel tasks, dispatches subtasks, monitors progress\n- **`/prodmasterai qa`** — 11-phase quality check with health scores across 8 categories, atomic fix commits\n- **`/prodmasterai cso`** — 14-phase OWASP + STRIDE security audit, exploit path required for every finding\n- **`/prodmasterai auto <goal>`** — fully autonomous: brainstorm → plan → implement → test → PR, no hand-holding\n\nAnd then it self-improves: every N tasks the plugin runs a convergence loop, detects skill gaps, and evolves its own skills. Your local improvements can be pushed upstream with `/prodmasterai update`.\n\n---\n\n## Platform Setup\n\n**Windows** — works out of the box (uses `run-hook.cmd` + PowerShell).\n\n**macOS / Linux** — after install, edit `hooks/hooks.json` and `.claude-plugin/hooks.json`, change the SessionStart command to:\n```json\n\"command\": \"\\\"${CLAUDE_PLUGIN_ROOT}/hooks/run-hook.sh\\\" session-start\"\n```\n\nPython 3 must be on PATH (default on macOS/Linux) — required for the safety hooks.\n\n---\n\n## Quick Start\n\n```\n/prodmasterai help\n```\n\nOr jump straight in:\n```\n/prodmasterai build user authentication\n```\n\nThe plugin reads your current state and routes to the right skill — no need to remember command names.\n\n---\n\n## All Commands\n\n| Command | What it does |\n|---|---|\n| `/prodmasterai` | Reads state and acts — or asks exactly one clarifying question |\n| `/prodmasterai build <feature>` | Breaks feature into tasks, coordinates parallel implementation |\n| `/prodmasterai pull latest` | `git pull`, repo health check, run tests |\n| `/prodmasterai should we A or B?` | Scored decision with deep reasoning |\n| `/prodmasterai report` | Prints full productivity report to terminal |\n| `/prodmasterai review` | Two-pass code review with live diff context, auto-fix mechanicals |\n| `/prodmasterai qa` | 11-phase QA, health score across 8 categories, atomic fix commits |\n| `/prodmasterai qa-only` | Findings-only QA, no fixes, screenshot evidence, baseline regression |\n| `/prodmasterai ship` | Completeness-principle pre-merge: tests → coverage → review → changelog → PR |\n| `/prodmasterai deploy` | Platform auto-detect, dry-run, canary verification, revert escape hatch |\n| `/prodmasterai cso` | 14-phase security audit (OWASP + STRIDE), exploit-path required for every finding |\n| `/prodmasterai dependency-audit` | CVE scan across npm/pip/bundler/go — blocks on CRITICAL CVEs |\n| `/prodmasterai secret-scan` | 25+ credential patterns, staged-file scan, git history check |\n| `/prodmasterai benchmark` | Core Web Vitals, bundle size, regression alerts, 4 modes |\n| `/prodmasterai codex` | Cross-model adversarial review, PASS/FAIL gate, cost tracking |\n| `/prodmasterai document-release` | Post-ship doc sync, CHANGELOG polish, consistency checks |\n| `/prodmasterai learn <topic>` | Research any topic, generate a production-ready SKILL.md |\n| `/prodmasterai queue add <goal>` | Add goal to sequential execution queue |\n| `/prodmasterai queue list` | Show pending/running/done queue |\n| `/prodmasterai queue run` | Run all queued tasks sequentially, auto-advance |\n| `/prodmasterai explore <goal>` | Run 2+ approaches in isolated worktrees, pick best by test pass rate |\n| `/prodmasterai auto <goal>` | Fully autonomous: brainstorm → plan → implement → test → PR |\n| `/prodmasterai resume` | Per-decision review and rollback of auto-pilot sessions |\n| `/prodmasterai plugins` | Show installed/available plugins, auto-install when needed |\n| `/prodmasterai cycle done — <summary>` | Record completed cycle, fire measure + learn |\n| `/prodmasterai update` | Push locally evolved improvements upstream via PR |\n| `/prodmasterai help` | Show all skills, triggers, and examples |\n| `/auto-pilot-revoke` | Stop a running auto-pilot, commit progress, reset lock |\n| `/evolve` | Convergence loop until all skills clean |\n\n---\n\n## Skills (31 total)\n\n| Skill | Purpose |\n|---|---|\n| `auto-pilot` | Fully autonomous unattended development session |\n| `auto-pilot-revoke` | Gracefully stop a running auto-pilot |\n| `benchmark` | Core Web Vitals, bundle size, 4 measurement modes |\n| `checkpoint` | Save and restore in-flight work across context resets |\n| `codex` | Cross-model adversarial code review |\n| `cso` | 14-phase security audit (OWASP + STRIDE) |\n| `decide` | Scored decision recommendation with deep reasoning |\n| `dependency-audit` | CVE scan across npm/pip/bundler/go |\n| `deploy` | Platform-aware deployment with dry-run and revert |\n| `dev-loop` | Repeating development loop for iterative tasks |\n| `document-release` | Post-ship doc sync and CHANGELOG polish |\n| `evolve-self` | Convergence loop — runs until all skills pass quality check |\n| `help` | Show all skills, triggers, and examples |\n| `learn` | Research a topic and generate a SKILL.md |\n| `measure` | Record cycle metrics: tasks, QA score, reviews, hours |\n| `orchestrate` | Break a feature into tracked parallel tasks |\n| `parallel-explore` | Run 2+ approaches in isolated worktrees, compare results |\n| `plugin-manager` | Detect, install, and manage plugins |\n| `prodmasterai` | Master entry point — routes to all other skills |\n| `qa` | 11-phase QA with health scoring and atomic fix commits |\n| `qa-only` | Findings-only QA with screenshot evidence |\n| `report` | Full productivity report printed to terminal |\n| `research-resolve` | Deep research to resolve a blocked task |\n| `resume` | Per-decision review and rollback of auto-pilot sessions |\n| `review` | Two-pass code review with auto-fix |\n| `secret-scan` | 25+ credential patterns, staged-file and history scan |\n| `ship` | Pre-merge completeness gate |\n| `skill-forge` | Create new production-ready skills from research |\n| `smooth-dev` | Git pull, repo health check, test run |\n| `task-queue` | Sequential task queue with auto-advance |\n| `token-efficiency` | Reduce token consumption and optimize context usage |\n\n---\n\n## Safety Hooks (4 active)\n\nThe plugin installs four hooks that run automatically in every session:\n\n| Hook | Fires on | What it does |\n|---|---|---|\n| `session-start` | Session open | Injects active features, patterns, gaps, open tasks, installed plugins |\n| `pre-tool-bash` | Every Bash call | Blocks: `rm -rf`, force push, `git reset --hard`, `DROP TABLE`, `chmod 777`, pipe-to-shell, base64-to-shell, PATH hijack, AWS key exports |\n| `post-tool-write` | Every Write/Edit | Scans for AWS/GitHub/OpenAI keys, private key material, SQL injection, XSS sinks, unsafe deserialization. Blocks on critical secrets. |\n| `stop-quality-gate` | Session stop | Blocks exit when: critical secrets leaked, critical CVEs unfixed, tests failing during ship/deploy |\n\nAll hooks are pure Python (standard library only, no pip installs). They fail open — a hook crash never silently blocks valid work.\n\n---\n\n## Optional Connectors\n\nAll inactive by default. Opt-in by editing the connector file:\n\n| Connector | How to activate |\n|---|---|\n| GitHub | Set `active: true` in `memory/connectors/github.md`, configure MCP server |\n| Slack | Set `active: true` in `memory/connectors/slack.md`, add webhook URL |\n| Linear | Set `active: true` in `memory/connectors/linear.md`, configure MCP server |\n| Superpowers | Set `active: true` in `memory/connectors/superpowers.md` |\n\nCredential files (`slack.md`, `linear.md`) are in `.gitignore` and never committed.\n\n---\n\n## External Network Access\n\nThe plugin makes no outbound network calls on its own. External access only occurs when:\n\n- A connector is explicitly activated and a Claude Code MCP server is configured by the user\n- A skill uses Claude Code's built-in `WebSearch` or `WebFetch` tools (user-approved per-call)\n- The `dependency-audit` skill queries public package registries to check CVE data\n\nNo telemetry. No analytics. No data leaves the machine unless you configure a connector.\n\n---\n\n## Memory and State\n\nAll state is stored locally in `memory/`. Nothing is sent upstream automatically.\n\n| File | Purpose |\n|---|---|\n| `memory/usage-log.md` | Per-invocation log, processed locally for auto-session measure |\n| `memory/skill-performance.md` | Velocity and quality metrics per cycle |\n| `memory/security-gate-state.json` | Active security gate state for the stop hook |\n| `memory/connectors/` | Connector configuration (most files git-ignored) |\n\n---\n\n## Requirements\n\n- Claude Code CLI (any recent version)\n- Python 3.8+ (standard library only — no pip installs required)\n- Git\n\n---\n\n## Installation\n\n```bash\n# From the plugin's own marketplace\nclaude plugin install prodmaster-ai@prodmaster-ai-marketplace\n\n# Verify\nclaude plugin list\n\n# Get started\n/prodmasterai help\n```\n\n---\n\n## Changelog\n\nSee [CHANGELOG.md](CHANGELOG.md) for version history.\n\n---\n\n## License\n\nMIT — see [LICENSE](LICENSE).\n\n**Author:** Sujith Krishnan — sujith.krishnan@techjays.com\n**Organization:** techjays (https://techjays.com)\n**Repository:** https://github.com/sujithkrishnan-dev/prodmaster-ai\n",
  "bytes": 9777,
  "sha": "2e7c24702ecc6b714b38ebfac76abb8cb92441cf2302a6b4245042477598de40",
  "repo_slug": "sujithkrishnan-dev/prodmaster-ai",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/plg_sujithkrishnan_dev_prodmaster_ai_prodmas_f3233ea7/readme"
}