Skip to content
Back to search

pro-workflow

MarketplaceWorks with Claude CodeGitHub

Claude Code learns from your corrections: self-correcting memory that compounds over 50+ sessions. Context engineering, parallel worktrees, agent teams, and 17 battle-tested skills.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/rohitg00-pro-workflow/badge.svg)](/listings/rohitg00-pro-workflow)

HTML

<a href="/listings/rohitg00-pro-workflow"><img src="/api/listings/rohitg00-pro-workflow/badge.svg" alt="findsafeskills safety rating" /></a>

Install

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

Pro Workflow is published in two places: the Claude Code plugin marketplace (native), and skills add (cross-agent installer). Other agents do not have first-class plugins yet &mdash; skills add installs the skill bundle into each agent's native skill directory.

Claude Code (native)

/plugin marketplace add rohitg00/pro-workflow
/plugin install pro-workflow@pro-workflow

Cursor, Codex, Copilot CLI, Droid, Gemini CLI, OpenCode, and 26 more (via skills add)

skills add installs the 41 skills + 23 commands into each agent's native skill directory.

npx skills add rohitg00/pro-workflow

Notes:

- Use `rohitg00/pro-workflow` (the GitHub form), not the bare name &mdash; `skills add` resolves providers from `owner/repo`, not marketplace slugs.

Works with these agents: `cursor`, `codex`, `gemini-cli`, `opencode`, `github-copilot`, `droid` (factory), `antigravity`, `amp`, `clawdbot`, `cline`, `codebuddy`, `commandcode`, `continue`, `crush`, `goose`, `kilo`, `kiro-cli`, `mcpjam`, `mux`, `neovate`, `openhands`, `pi`, `qoder`, `qwen`, `roo`, `trae`, `universal`, `vercel`, `windsurf`, `zencoder`.

Roster notes (2026): `windsurf` relaunched as Devin Desktop, which speaks the Agent Client Protocol (ACP); `gemini-cli` was superseded by the Antigravity CLI for consumers. The adapter names above still resolve, but target the current product. MCP works across every agent in this list with the same setup. See [`references/modern-workflows-2026.md`](references/modern-workflows-2026.md) for the interop picture.

After install, run `skills sync` to register the skills with the target agent's config.

<details> <summary>Manual install (any agent, any OS)</summary>

If neither path works for your setup, clone and copy the bundle directly. Adjust the destination to your agent's skill directory (e.g. `~/.cursor/rules/`, `~/.gemini/extensions/`, etc.).

git clone https://github.com/rohitg00/pro-workflow.git /tmp/pw
cd /tmp/pw && npm install && npm run build

cp -r /tmp/pw/templates/split-claude-md/* ./.claude/
cp -r /tmp/pw/skills    ~/.claude/skills/
cp -r /tmp/pw/commands  ~/.claude/commands/
cp    /tmp/pw/hooks/hooks.json ~/.claude/hooks.json

</details>

First-run smoke test

/doctor              # confirms SQLite store, hooks, skills load
/wrap-up             # runs the end-of-session ritual (no-op on fresh install)

If `/doctor` reports `KB: missing`, run `cd ~/.claude/plugins/*/pro-workflow && npm install && npm run build` &mdash; the SQLite components need a build step a handful of marketplaces skip.

---

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

.claude-plugin/marketplace.json · Manifest · 48 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. {
  2. "name": "pro-workflow",
  3. "owner": {
  4. "name": "Rohit Ghumare",
  5. "email": "ghumare64@gmail.com"
  6. },
  7. "metadata": {
  8. "description": "Battle-tested Claude Code workflows from power users"
  9. },
  10. "plugins": [
  11. {
  12. "name": "pro-workflow",
  13. "source": "./",
  14. "description": "Complete AI coding workflow: context engineering, agent teams, 41 skills, 8 agents, 23 commands, 13 guides, cross-agent support, and searchable learnings.",
  15. "author": {
  16. "name": "Rohit Ghumare"
  17. },
  18. "homepage": "https://github.com/rohitg00/pro-workflow",
  19. "repository": "https://github.com/rohitg00/pro-workflow",
  20. "license": "MIT",
  21. "keywords": [
  22. "workflow",
  23. "productivity",
  24. "self-correction",
  25. "worktrees",
  26. "hooks",
  27. "agents",
  28. "orchestration",
  29. "cross-agent",
  30. "cursor",
  31. "codex"
  32. ],
  33. "category": "workflow",
  34. "tags": [
  35. "workflow",
  36. "hooks",
  37. "agents",
  38. "commands",
  39. "productivity",
  40. "best-practices",
  41. "orchestration",
  42. "settings-guide",
  43. "cli-reference",
  44. "cross-agent"
  45. ]
  46. }
  47. ]
  48. }
README.md · README · 522 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. <p align="center">
  2. <img src="assets/banner.svg" alt="Pro Workflow" width="100%"/>
  3. </p>
  4. <p align="center">
  5. <a href="https://github.com/rohitg00/pro-workflow/stargazers"><img src="https://img.shields.io/github/stars/rohitg00/pro-workflow?style=for-the-badge&logo=github&color=D97757&labelColor=1e1e2e" alt="Stars"/></a>
  6. <a href="https://www.npmjs.com/package/pro-workflow"><img src="https://img.shields.io/npm/v/pro-workflow?style=for-the-badge&logo=npm&color=E8926F&labelColor=1e1e2e" alt="npm"/></a>
  7. <a href="https://github.com/rohitg00/pro-workflow/blob/main/LICENSE"><img src="https://img.shields.io/badge/license-MIT-22c55e?style=for-the-badge&labelColor=1e1e2e" alt="License"/></a>
  8. <a href="https://skills.sh"><img src="https://img.shields.io/badge/skills%20add-32%2B%20agents-f59e0b?style=for-the-badge&labelColor=1e1e2e" alt="skills add"/></a>
  9. <a href="https://rohitg00-pro-workflow.mintlify.app/"><img src="https://img.shields.io/badge/Docs-Mintlify-0D9373?style=for-the-badge&logo=mintlify&labelColor=1e1e2e" alt="Docs"/></a>
  10. </p>
  11. <h3 align="center">Your Claude Code gets smarter every session.</h3>
  12. <p align="center">
  13. Self-correcting memory + persistent FTS5-indexed wikis + auto-research loop, all on one SQLite store.<br/>
  14. Correct Claude once &mdash; it never repeats the mistake. Build a wiki on a topic &mdash; it grows itself overnight.<br/>
  15. <b>41 skills</b> &bull; <b>8 agents</b> &bull; <b>23 commands</b> &bull; <b>33 hook scripts across 22 events</b><br/>
  16. Works with <b>Claude Code</b>, <b>Cursor</b>, and <b>32+ agents</b> via skills add.
  17. </p>
  18. ---
  19. ## The Problem
  20. You correct Claude the same way 50 times. You explain conventions every new session. Context compacts, learnings vanish, mistakes repeat. You research the same topic in three different sessions because there is nowhere durable for the answers to land.
  21. **Every Claude Code user hits this wall.**
  22. ## The Solution
  23. Pro Workflow puts a single SQLite store underneath every session.
  24. - **Self-correction memory** &mdash; every correction becomes a rule, FTS5-searchable, auto-loaded on session start.
  25. - **Knowledge plane** &mdash; persistent research wikis on disk + FTS5 shadow index, queryable from any session, optionally grown by an auto-research loop.
  26. - **Quality gates** &mdash; LLM-powered hooks, deterministic git/secret guards, compaction-aware state, cost tracking.
  27. After 50 sessions you barely correct anything. After a week of auto-research, your wiki on a topic is denser than the curated lists you started from.
  28. <p align="center">
  29. <img src="assets/self-correction-demo.svg" alt="Self-Correction Loop" width="700"/>
  30. </p>
  31. ```
  32. Session 1: You → "Don't mock the database in tests"
  33. Claude → Proposes rule → You approve → Saved to SQLite
  34. Session 2: SessionStart loads all learnings + lists your wikis
  35. UserPromptSubmit auto-injects top wiki hits when relevant
  36. Claude writes integration tests, cites the right wiki page
  37. Session 50: Correction rate near zero. Wiki has 200 cited claims.
  38. ```
  39. ---
  40. ## Install
  41. Pro Workflow is published in two places: the Claude Code plugin marketplace (native), and skills add (cross-agent installer). Other agents do not have first-class plugins yet &mdash; skills add installs the skill bundle into each agent's native skill directory.
  42. ### Claude Code (native)
  43. ```bash
  44. /plugin marketplace add rohitg00/pro-workflow
  45. /plugin install pro-workflow@pro-workflow
  46. ```
  47. ### Cursor, Codex, Copilot CLI, Droid, Gemini CLI, OpenCode, and 26 more (via skills add)
  48. skills add installs the 41 skills + 23 commands into each agent's native skill directory.
  49. ```bash
  50. npx skills add rohitg00/pro-workflow
  51. ```
  52. Notes:
  53. - Use `rohitg00/pro-workflow` (the GitHub form), not the bare name &mdash; `skills add` resolves providers from `owner/repo`, not marketplace slugs.
  54. Works with these agents: `cursor`, `codex`, `gemini-cli`, `opencode`, `github-copilot`, `droid` (factory), `antigravity`, `amp`, `clawdbot`, `cline`, `codebuddy`, `commandcode`, `continue`, `crush`, `goose`, `kilo`, `kiro-cli`, `mcpjam`, `mux`, `neovate`, `openhands`, `pi`, `qoder`, `qwen`, `roo`, `trae`, `universal`, `vercel`, `windsurf`, `zencoder`.
  55. Roster notes (2026): `windsurf` relaunched as Devin Desktop, which speaks the Agent Client Protocol (ACP); `gemini-cli` was superseded by the Antigravity CLI for consumers. The adapter names above still resolve, but target the current product. MCP works across every agent in this list with the same setup. See [`references/modern-workflows-2026.md`](references/modern-workflows-2026.md) for the interop picture.
  56. After install, run `skills sync` to register the skills with the target agent's config.
  57. <details>
  58. <summary>Manual install (any agent, any OS)</summary>
  59. If neither path works for your setup, clone and copy the bundle directly. Adjust the destination to your agent's skill directory (e.g. `~/.cursor/rules/`, `~/.gemini/extensions/`, etc.).
  60. ```bash
  61. git clone https://github.com/rohitg00/pro-workflow.git /tmp/pw
  62. cd /tmp/pw && npm install && npm run build
  63. cp -r /tmp/pw/templates/split-claude-md/* ./.claude/
  64. cp -r /tmp/pw/skills ~/.claude/skills/
  65. cp -r /tmp/pw/commands ~/.claude/commands/
  66. cp /tmp/pw/hooks/hooks.json ~/.claude/hooks.json
  67. ```
  68. </details>
  69. ### First-run smoke test
  70. ```bash
  71. /doctor # confirms SQLite store, hooks, skills load
  72. /wrap-up # runs the end-of-session ritual (no-op on fresh install)
  73. ```
  74. If `/doctor` reports `KB: missing`, run `cd ~/.claude/plugins/*/pro-workflow && npm install && npm run build` &mdash; the SQLite components need a build step a handful of marketplaces skip.
  75. ---
  76. ## What to type first
  77. After install you have **41 skills** and **23 slash commands**. You don't need to memorize them; the agent picks the right skill from your prompt. The five commands below cover 80% of daily use:
  78. | When | Command | What it does |
  79. |---|---|---|
  80. | **Wrong correction repeats** | `/learn-rule` | Capture the correction as a rule. Loaded on every future `SessionStart`. |
  81. | **End of a coding session** | `/wrap-up` | Audit changes, persist learnings, write a handoff doc. |
  82. | **Researching a topic** | `/wiki init <slug>` | Spin up a persistent FTS5 wiki. Auto-injected when you mention the topic later. |
  83. | **Stuck on a hard bug** | `/develop` | Research &rarr; Plan &rarr; Implement phases with validation gates. |
  84. | **Before a PR** | `/smart-commit` | Quality gates, staged review, conventional commit message. |
  85. Full list: [`commands/`](./commands) &middot; [`skills/`](./skills) &middot; [`/list`](./commands/list.md) inside any session.
  86. ---
  87. ## 60-second tour
  88. ```bash
  89. # 1. Self-correction (existing)
  90. /learn-rule # capture a correction
  91. /wrap-up # end session, persist learnings, audit changes
  92. /insights # heatmaps, trends, productivity
  93. # 2. Knowledge plane (v3.3, new)
  94. /wiki init agent-memory --title "Agent Memory" --flavor research
  95. /wiki page agent-memory wiki/concepts/episodic-memory.md --type concept
  96. /wiki ask "what is episodic memory" --wiki agent-memory
  97. # 3. Auto-research (budget-capped, opt-in)
  98. /wiki seed agent-memory "memory consolidation in agents"
  99. /wiki research agent-memory --max-pages 5 --budget-usd 0.50
  100. # 4. Hybrid retrieval (BM25 + vector RRF, optional)
  101. /wiki embed agent-memory # configure an OpenAI or Voyage key for this plugin
  102. /wiki hybrid "consolidation patterns" --wiki agent-memory
  103. # 5. Multi-LLM deliberation (transcript persists as a wiki page)
  104. /wiki council "should we adopt episodic memory?" --wiki agent-memory
  105. # 6. Browse the wiki visually (single-file HTML, S3-shareable)
  106. /wiki view agent-memory
  107. open ~/.pro-workflow/wikis/agent-memory/derived/viewer.html
  108. # Kill switch for any auto loop
  109. touch ~/.pro-workflow/STOP
  110. ```
  111. `UserPromptSubmit` auto-loads top-3 wiki hits when prompts mention indexed topics. `SessionStart` lists registered wikis and recent learnings.
  112. ---
  113. ## What's new in v3.3
  114. Persistent knowledge plane on top of self-correction memory.
  115. | Skill | Purpose |
  116. |-------|---------|
  117. | **wiki-builder** | Persistent FTS5-indexed research wikis. 9 flavors: research, paper, domain, product, person, organization, project, codebase, incident. Path-traversal-guarded. |
  118. | **wiki-query** | BM25 retrieval with snippets. `ask`, `related`, `show`. Auto-injects on `UserPromptSubmit`. |
  119. | **wiki-research-loop** | Budget-capped BFS. Pluggable source fetchers (web/arXiv/GitHub + custom). Convergence detection, kill-switch, atomic seed claim, try/finally state guards. |
  120. | **llm-council** | Provider-agnostic 3-phase deliberation (Anthropic/OpenAI/OpenRouter/Fireworks/custom). `Promise.allSettled` so one provider failure doesn't abort the run. Transcript persists as a wiki page. |
  121. | **survey-generator** | Provider-agnostic literature survey. Output target = wiki markdown page. Bibliography validation (uniqueness + section-paper refs), citation IDs aligned with `sources.md` rows. |
  122. | **wiki-viewer** | Single-file HTML viewer for any wiki: pages + sources + seeds + link graph + in-browser search + "copy as seed" CTAs. S3-shareable. Applied lessons from [Thariq Shihipar's HTML-as-output thesis](https://x.com/trq212/status/2034017024445244382). |
  123. Plus: `/wiki` command (now with `view`), `learn-rule` `Wiki: <slug>` scoping, schema additions (`wikis`, `wiki_pages` + FTS5, `wiki_sources`, `wiki_claims`, `wiki_seeds`, `wiki_embeddings`, `learnings_wiki`), reactive file-watcher seed enqueue, cron-tick driver, `/doctor` extended with KB + provider sections.
  124. ---
  125. ## Comparison
  126. | Feature | Pro Workflow | [Superpowers](https://github.com/obra/superpowers) | [ECC](https://github.com/affaan-m/everything-claude-code) | [gstack](https://github.com/garrytan/gstack) | [GSD](https://github.com/gsd-build/get-shit-done) |
  127. |---------|:-----------:|:-----------:|:---:|:------:|:---:|
  128. | Self-correcting memory (SQLite + FTS5) | **Yes** | No | No | No | No |
  129. | Persistent research wikis (FTS5) | **Yes** | No | No | No | No |
  130. | Auto-research loop (budget-capped BFS) | **Yes** | No | No | No | No |
  131. | Hybrid retrieval (BM25 + vector + RRF) | **Yes** | No | No | No | No |
  132. | Multi-provider LLM council | **Yes** | No | No | No | No |
  133. | LLM-powered hooks (`type: "prompt"`) | **Yes** | No | No | No | No |
  134. | Permission denial analysis | **Yes** | No | No | No | No |
  135. | Compaction-aware state preservation | **Yes** | No | No | No | No |
  136. | Cost tracking and budget alerts | **Yes** | No | No | No | No |
  137. | MCP overhead auditing | **Yes** | No | No | No | No |
  138. | Cross-agent (32+ agents via skills add) | **Yes** | No | Some | No | No |
  139. | Skills | 34 | 14 | 140+ | 18+ | 0 |
  140. | Agents | 8 | 5 | 36 | 0 | 18 |
  141. | Commands | 22 | 3 | 60+ | 5+ | 57 |
  142. | Hook events | 22 | 8 | 18 | 0 | 0 |
  143. ---
  144. ## What's inside
  145. ### 41 skills
  146. **Knowledge plane (new in v3.3)**
  147. | Skill | What it does |
  148. |-------|--------------|
  149. | **wiki-builder** | Scaffold + register FTS5-indexed research wikis |
  150. | **wiki-query** | BM25 retrieval, snippets, related, show |
  151. | **wiki-research-loop** | Budget-capped BFS over web/arXiv/GitHub fetchers |
  152. | **llm-council** | Provider-agnostic 3-phase multi-LLM deliberation |
  153. | **survey-generator** | Literature survey artifact, output to a wiki page |
  154. | **wiki-viewer** | Single-file HTML viewer (pages, sources, seeds, link graph, search) |
  155. **Quality gates and observability**
  156. | Skill | What it does |
  157. |-------|--------------|
  158. | **smart-commit** | Quality gates, staged review, conventional commits |
  159. | **llm-gate** | AI-powered commit and secret hooks (`type: "prompt"`) |
  160. | **permission-tuner** | Analyze denials, generate allow/deny rules |
  161. | **compact-guard** | State preservation through compaction cycles |
  162. | **cost-tracker** | Session cost awareness with budget benchmarks |
  163. | **mcp-audit** | MCP server token-overhead analysis |
  164. | **token-efficiency** | Anti-sycophancy + tool-call budgets + read-before-write |
  165. | **safe-mode** | Guardrails for destructive operations |
  166. | **insights** | Session analytics, correction trends, productivity |
  167. | **thoroughness-scoring** | Rate completeness of implementations |
  168. | **deslop** | Remove AI-generated code slop from a diff |
  169. **Memory and learning**
  170. | Skill | What it does |
  171. |-------|--------------|
  172. | **pro-workflow** | Core 8 patterns |
  173. | **learn-rule** | Capture corrections (now wiki-scopeable) |
  174. | **replay-learnings** | Surface past learnings for the current task |
  175. | **wrap-up** | End-of-session ritual |
  176. | **session-handoff** | Resume documents for the next session |
  177. **Orchestration and engineering loop**
  178. | Skill | What it does |
  179. |-------|--------------|
  180. | **orchestrate** | Multi-phase Research → Plan → Implement → Review |
  181. | **agent-teams** | Multi-instance coordination, shared task list |
  182. | **batch-orchestration** | Parallel worktree agents for large changes |
  183. | **parallel-worktrees** | Git worktree setup for zero dead time |
  184. | **context-engineering** | Write/Select/Compress/Isolate framework |
  185. | **context-optimizer** | Token management, context budget, MCP audit |
  186. | **auto-setup** | Auto-detect project type, configure quality gates |
  187. | **file-watcher** | Reactive workflows on config and dependency changes |
  188. | **bug-capture** | Capture defects as durable issues without leaking paths |
  189. | **module-map** | One-screen map of an unfamiliar codebase area |
  190. | **plan-interrogate** | Stress-test a plan by walking its decision tree |
  191. | **sprint-status** | Track multi-session progress |
  192. ### 8 agents
  193. | Agent | Purpose |
  194. |-------|---------|
  195. | **planner** | Break down complex tasks (read-only, approval-gated) |
  196. | **reviewer** | Code review and security audit (checklist-based) |
  197. | **scout** | Confidence-gated exploration (background, worktree-isolated) |
  198. | **orchestrator** | Multi-phase feature development |
  199. | **debugger** | Systematic, hypothesis-driven bug investigation |
  200. | **context-engineer** | Context-window analysis (lightweight, read-only) |
  201. | **permission-analyst** | Denial-pattern analysis, rule-optimization recommendations |
  202. | **cost-analyst** | Token-usage analysis, identify expensive operations |
  203. ### 23 commands
  204. | Command | What it does |
  205. |---------|--------------|
  206. | `/wiki` | Knowledge-plane entry: init, page, ask, hybrid, seed, research, council, survey, embed, status |
  207. | `/develop` | Multi-phase feature build with validation gates |
  208. | `/commit` | Smart commit with quality gates |
  209. | `/wrap-up` | End-of-session checklist and learning capture |
  210. | `/learn-rule` / `/learn` | Extract correction or learn Claude Code best practices |
  211. | `/doctor` | Setup health check (now includes wiki KB + council providers) |
  212. | `/insights` | Session analytics and correction heatmap |
  213. | `/replay` | Surface past learnings for the current task |
  214. | `/handoff` | Session handoff document |
  215. | `/search` / `/list` | Search and list stored learnings |
  216. | `/deslop` | Remove AI-generated code slop |
  217. | `/context-optimizer` | Audit and optimize context-window usage |
  218. | `/parallel` | Set up git worktrees for parallel sessions |
  219. | `/safe-mode` | Toggle destructive-operation guardrails |
  220. | `/sprint-status` | Track multi-session progress |
  221. | `/auto-setup` | Auto-detect project type, configure quality gates |
  222. | `/compact-guard` | Protected compaction with state preservation |
  223. | `/cost-tracker` | Track session costs and optimization tips |
  224. | `/mcp-audit` | Audit MCP servers for token overhead |
  225. | `/permission-tuner` | Generate allow/deny rules from denial patterns |
  226. ### 33 hook scripts across 22 events
  227. `SessionStart`, `SessionEnd`, `UserPromptSubmit`, `PreToolUse`, `PostToolUse`, `Stop`, `PreCompact`, `PostCompact`, `SubagentStart`, `SubagentStop`, `TaskCreated`, `TaskCompleted`, `PermissionRequest`, `PermissionDenied`, `PostToolUseFailure`, `TeammateIdle`, `StopFailure`, `FileChanged`, `ConfigChange`, `Notification`, `Setup`, `CwdChanged`.
  228. Selected high-leverage hooks:
  229. | Hook | Script | Effect |
  230. |------|--------|--------|
  231. | `SessionStart` | `session-start.js` | Loads learnings; lists registered wikis |
  232. | `UserPromptSubmit` | `prompt-submit.js` | Auto-injects top-3 wiki hits when prompts match the index |
  233. | `Stop` | `learn-capture.js` | Auto-captures `[LEARN]` blocks (now parses `Wiki: <slug>`) |
  234. | `FileChanged` | `file-changed.js` | Edits inside a wiki tree enqueue verify-seeds |
  235. | `PreToolUse(Bash)` | `commit-validate.js`, `git-blast-radius.js`, `pre-push-check.js` | Conventional commit + destructive op + push guardrails |
  236. | `PreToolUse(Write)` | `secret-scan.js` | LLM-powered secret detection |
  237. | `PreCompact` / `PostCompact` | `pre-compact.js`, `post-compact.js` | Save and re-inject critical context summary |
  238. | `PreToolUse(Read)`, `PostToolUse(Read)` | `reread-tracker.js` | Records each completed Read and tells Claude when an unchanged file is read again; set `reread_tracker.block: true` in `config.json` or `PRO_WORKFLOW_REREAD_BLOCK=1` to block the read |
  239. ### Reference guides
  240. | Guide | What's covered |
  241. |-------|----------------|
  242. | [`modern-workflows-2026.md`](references/modern-workflows-2026.md) | Multi-agent orchestration, Agent Teams, checkpoints, ACP/MCP interop, skills-commands merger |
  243. | [`../references/models-2026.md`](references/models-2026.md) | Current model tiers, effort routing, task budgets |
  244. | [`settings-guide.md`](references/settings-guide.md) | Settings keys, permission modes, hierarchy |
  245. | [`cli-cheatsheet.md`](references/cli-cheatsheet.md) | CLI flags, keyboard shortcuts, slash commands |
  246. | [`orchestration-patterns.md`](references/orchestration-patterns.md) | Command → Agent → Skill architecture, agent teams |
  247. | [`context-engineering.md`](references/context-engineering.md) | Write/Select/Compress/Isolate, memory taxonomy, compaction |
  248. | [`agent-teams.md`](references/agent-teams.md) | Setup, decomposition, teams vs sub-agents |
  249. | [`context-loading.md`](references/context-loading.md) | CLAUDE.md loading, agent memory, skills discovery |
  250. | [`cross-agent-workflows.md`](references/cross-agent-workflows.md) | Claude Code + Cursor together, skills add cross-agent |
  251. | [`decision-framework.md`](references/decision-framework.md) | When to use which pattern |
  252. | [`daily-habits.md`](references/daily-habits.md) | Session habits, debugging tips, anti-patterns |
  253. | [`new-features.md`](references/new-features.md) | Checkpoints, remote sessions, agent and prompt hooks |
  254. | [`claude-code-resources.md`](references/claude-code-resources.md) | Official docs map and resource index |
  255. ---
  256. ## How it works
  257. ### `/develop` flow
  258. <p align="center">
  259. <img src="assets/workflow-flow.svg" alt="Development Flow" width="900"/>
  260. </p>
  261. Research before planning. Plan before implementing. Review before committing. Validation gates between every phase.
  262. ### Architecture
  263. <p align="center">
  264. <img src="assets/architecture.svg" alt="Architecture Diagram" width="900"/>
  265. </p>
  266. Command → Agent → Skill, layered over a single SQLite store. The knowledge plane (v3.3) plugs into the same store as the learning rules.
  267. ### Storage
  268. ```
  269. ~/.pro-workflow/
  270. ├── data.db # learnings, sessions, wikis (registry), wiki_pages (+FTS5),
  271. │ # wiki_sources, wiki_claims, wiki_seeds, wiki_embeddings,
  272. │ # learnings_wiki
  273. ├── wikis/<slug>/ # global-scope wikis (default location)
  274. ├── council/<session-id>/ # llm-council transcripts
  275. ├── fetchers/ # user-supplied custom source fetchers
  276. ├── tick.log # cron-driven research-tick log
  277. └── STOP # touch this file to halt every research loop
  278. ```
  279. Project-scope wikis live at `<project>/.claude/wikis/<slug>/` and are committable.
  280. ---
  281. ## Patterns
  282. | Pattern | What it does |
  283. |---------|--------------|
  284. | **Self-Correction Loop** | Claude learns from corrections automatically |
  285. | **Knowledge Plane** | Wiki + auto-research loop on the same SQLite store |
  286. | **Multi-Phase Dev** | Research → Plan → Implement → Review with gates |
  287. | **Parallel Worktrees** | Zero dead time with native `claude -w` |
  288. | **Wrap-Up Ritual** | End sessions with intention, capture learnings |
  289. | **Split Memory** | Modular CLAUDE.md for complex projects |
  290. | **80/20 Review** | Batch reviews at checkpoints |
  291. | **Context Engineering** | Write/Select/Compress/Isolate for token management |
  292. | **Agent Teams** | Multi-instance coordination with shared task list |
  293. | **Batch Orchestration** | Parallel worktree agents for large-scale changes |
  294. | **LLM Gates** | AI-powered verification before destructive operations |
  295. | **Permission Tuning** | Denial-pattern analysis to reduce prompt fatigue |
  296. | **Token Efficiency** | Anti-sycophancy + tool-call budgets + read-before-write |
  297. | **Multi-LLM Council** | Provider-agnostic 3-phase deliberation for high-stakes calls |
  298. ---
  299. ## Cross-agent support
  300. Works across 32+ AI coding agents via [skills add](https://skills.sh):
  301. ```bash
  302. npx skills add rohitg00/pro-workflow
  303. ```
  304. Supported: Claude Code, Cursor, Codex CLI, Gemini CLI, Windsurf, OpenCode, Kiro, Amp, Goose, Roo, and 27 more.
  305. ---
  306. ## Configuration
  307. ### Settings
  308. See [`settings.example.json`](settings.example.json) for production-ready configuration: permission rules, an `autoMode` block, sandbox settings, output style, auto-compaction, custom spinner verbs. The classifier reads `autoMode` only from `~/.claude/settings.json`, so put that block there.
  309. ### MCP
  310. See [`mcp-config.example.json`](mcp-config.example.json):
  311. - **context7** &mdash; live documentation lookup
  312. - **playwright** &mdash; browser automation (most token-efficient)
  313. - **GitHub** &mdash; PRs, issues, code search
  314. Rule: start with three MCPs, add only for concrete needs.
  315. ### Provider credentials
  316. Configure optional API keys in the plugin configuration dialog. Sensitive values stay in Claude Code's secure credential storage and reach API-backed runners through the bundled `providers` MCP server. Ordinary environment keys such as `OPENAI_API_KEY` and `GH_TOKEN` are no longer consumed automatically.
  317. Standalone CLI users must explicitly supply the corresponding `PRO_WORKFLOW_*` variable below. Local BM25 search, learnings, and deterministic guards require no credentials. See [provider configuration and migration](references/provider-configuration.md) for setup and MCP examples.
  318. ### Knowledge plane env
  319. | Env | When |
  320. |-----|------|
  321. | `WIKI_ROOT` | Override default `~/.pro-workflow/wikis` |
  322. | `PRO_WORKFLOW_OPENAI_API_KEY` / `PRO_WORKFLOW_VOYAGE_API_KEY` | Explicit standalone embedding credentials |
  323. | `PRO_WORKFLOW_ANTHROPIC_API_KEY` / `PRO_WORKFLOW_OPENAI_API_KEY` / `PRO_WORKFLOW_OPENROUTER_API_KEY` / `PRO_WORKFLOW_FIREWORKS_API_KEY` / `LLM_COUNCIL_BASE_URL`+`PRO_WORKFLOW_LLM_COUNCIL_API_KEY` | Explicit standalone council and survey credentials; the four named providers also support optimization |
  324. | `WIKI_LOOP_BUDGET_USD` / `WIKI_LOOP_MAX_PAGES` / `WIKI_LOOP_MAX_DEPTH` | Per-run loop overrides |
  325. | `PRO_WORKFLOW_GITHUB_TOKEN` | Explicit standalone GitHub research token; public requests work without it |
  326. ### Optional: System 1 classifiers
  327. Off by default. Set `PRO_WORKFLOW_SYSTEM_ONE=laya` (local Laya server) or `=jev` (with a TypeSafe key in plugin configuration, or `PRO_WORKFLOW_TYPESAFE_API_KEY` for standalone use) to add fast correction hints and risk suggestions. The classifier never allows or denies a tool call. See [`references/system-one-classifiers.md`](references/system-one-classifiers.md).
  328. ---
  329. ## Structure
  330. ```text
  331. pro-workflow/
  332. ├── skills/ # 41 skills
  333. ├── agents/ # 8 agents
  334. ├── commands/ # 23 slash commands
  335. ├── scripts/ # 33 hook scripts (22 events)
  336. ├── references/ # workflow, model, and skill reference guides
  337. ├── docs/ # GitHub Pages infographic
  338. ├── rules/ # rule packs (Cursor + universal)
  339. ├── contexts/ # context modes
  340. ├── templates/ # split CLAUDE.md + AGENTS.md
  341. ├── src/ # TypeScript source (SQLite + embeddings)
  342. └── config.json
  343. ```
  344. ---
  345. ## Tips from the community
  346. > "80% of my code is written by AI, 20% is spent reviewing and correcting it." &mdash; [Andrej Karpathy](https://x.com/karpathy/status/2015883857489522876)
  347. > "Use subagents to throw more compute at a problem &mdash; offload tasks to keep your main context clean." &mdash; [Boris Cherny](https://x.com/bcherny/status/2017742755737555434)
  348. > "If you do something more than once a day, turn it into a skill or command." &mdash; [Boris Cherny](https://x.com/bcherny/status/2017742748984742078)
  349. > "Write detailed specs and reduce ambiguity before handing work off." &mdash; [Boris Cherny](https://x.com/bcherny/status/2017742752566632544)
  350. > "Skill description field is a trigger, not a summary &mdash; write it for the model." &mdash; [Thariq Shihipar](https://x.com/trq212/status/2033949937936085378)
  351. ---
  352. ## Philosophy
  353. 1. **Compound improvements** &mdash; small corrections lead to big gains.
  354. 2. **Trust but verify** &mdash; let AI work, review at checkpoints.
  355. 3. **Zero dead time** &mdash; parallel sessions keep momentum.
  356. 4. **Memory is precious** &mdash; both yours and Claude's.
  357. 5. **Persistent over ephemeral** &mdash; if you'll need it next session, write it down on disk.
  358. 6. **Orchestrate, don't micromanage** &mdash; wire patterns together, let agents execute.
  359. ---
  360. ## Related projects
  361. | Project | Description |
  362. |---------|-------------|
  363. | [everything-claude-code](https://github.com/affaan-m/everything-claude-code) | 140+ skills, 36 agents — the comprehensive collection |
  364. | [claude-code-best-practice](https://github.com/shanraisshan/claude-code-best-practice) | Curated tips, comparisons, community best practices |
  365. | [skills](https://skills.sh) | Registry and CLI to add skills across 32+ agents |
  366. | [awesome-claude-code-toolkit](https://github.com/rohitg00/awesome-claude-code-toolkit) | 850+ files, 135 agents, 176 plugins — curated directory |
  367. ---
  368. ## Star history
  369. <a href="https://star-history.com/#rohitg00/pro-workflow&Date">
  370. <picture>
  371. <source media="(prefers-color-scheme: dark)" srcset="https://api.star-history.com/svg?repos=rohitg00/pro-workflow&type=Date&theme=dark" />
  372. <source media="(prefers-color-scheme: light)" srcset="https://api.star-history.com/svg?repos=rohitg00/pro-workflow&type=Date" />
  373. <img alt="Star History Chart" src="https://api.star-history.com/svg?repos=rohitg00/pro-workflow&type=Date" />
  374. </picture>
  375. </a>
  376. ---
  377. <p align="center">
  378. <br/>
  379. <b>If pro-workflow saves you time, star the repo so others can find it.</b>
  380. <br/><br/>
  381. <a href="https://github.com/rohitg00/pro-workflow/stargazers"><img src="https://img.shields.io/github/stars/rohitg00/pro-workflow?style=for-the-badge&logo=github&color=D97757&labelColor=1e1e2e" alt="Stars"/></a>
  382. <br/><br/>
  383. <a href="https://rohitg00-pro-workflow.mintlify.app/">Documentation</a> &bull;
  384. <a href="https://skills.sh">Skills Registry</a> &bull;
  385. <a href="https://github.com/rohitg00/pro-workflow/issues">Report Issues</a> &bull;
  386. <a href="references/">Reference Guides</a>
  387. </p>

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.