Centrally manage agents, skills, MCP servers, and AI guidelines for your whole team. Plan work in the dashboard, execute it with your AI agent in one click.
$ npx @cgdevs/pigi sync
✓ Written CLAUDE.md
✓ Written .cursor/rules
✓ Written .github/copilot-instructions.md
✓ Written AGENTS.md
✓ Written .windsurf/rules.md
✓ Written GEMINI.md
6 AI tools synced in 0.4s
Every dev runs slightly different agents, skills, and MCP servers — leading to inconsistent AI behavior across the team.
New developers waste hours setting up each AI tool from scratch, copying configs from teammates who may already be out of date.
Agents, skills, and config changes live in Slack messages or internal wikis. There's no single place to see what AI stack your team actually runs.
When you add a new agent, publish a new skill, or update AI guidelines, good luck getting everyone to actually apply the change.
Pigi is a central control plane for your team's entire agentic AI stack. Define agents, skills, and configs once — every developer stays up to date automatically.
Get started freeDefine your agentic stack once
Set up agents, skills, MCP servers, and AI guidelines in one place. Bundle them into profiles, assign profiles to projects.
Developers sync with one command
Run npx @cgdevs/pigi sync and every AI tool gets its native config written to disk.
Git hook keeps it automatic
The optional post-merge hook syncs on every git pull. Config changes propagate to the whole team instantly.
Built for engineering teams that take AI tooling seriously.
Central agentic stack management
Define agents, skills, MCP servers, and AI guidelines once in the dashboard. No more per-developer, per-tool setup.
One-command sync
Run npx @cgdevs/pigi sync and every AI tool gets its native config — agents, skills, MCPs, and rules — written to disk. Works with Claude, Cursor, Copilot, Codex, Gemini, and Windsurf.
Git hook automation
The optional post-merge hook syncs your entire agentic stack on every git pull. All tools updated, zero manual steps.
Profiles & projects
Bundle agents, skills, MCPs, and guidelines into named profiles. Assign different profiles to different projects with per-project overrides.
Secret references
Reference env-var names in configs. The CLI resolves values from each developer's local environment — secrets never leave their machine.
Role-based access
Owner, admin, and member roles with a full audit log. You control who can change your team's shared agentic configuration.
Marketplace
Browse and install community-built agents, skills, and MCP server configs in one click. Publish your own to share with the ecosystem.
AI-generated documents
Create shared context documents — guidelines, coding standards, project context — with built-in AI generation. Every tool reads the same rules on every sync.
Execute Plan
Define phases and specs in the dashboard, then hit Execute Plan in the desktop app. Your chosen AI agent picks up every pending phase and runs it — live, in a terminal.
Build your agentic stack in the dashboard
Define agents, skills, MCP servers, and AI guidelines. Bundle them into profiles and assign profiles to projects.
Developers sync with one command
Each developer runs the CLI once. It pulls agents, skills, MCPs, and AI rules and writes each AI tool's native config files to disk.
npx @cgdevs/pigi syncStay in sync automatically
The optional git hook runs pigi sync on every git pull. Your entire agentic stack propagates to every AI tool across the whole team instantly.
AI tools are only as good as the configs behind them. When every developer runs a slightly different setup, you don't just get inconsistency — you get unpredictable AI behavior at the worst possible times.
Pigi treats AI configuration the same way great teams treat code: versioned, centralized, and automatically deployed. One source of truth. Zero drift.
Define phases and specs in the dashboard. Open the desktop app, hit Execute Plan, and your chosen AI agent picks up every pending phase and runs it — live, in a terminal. Claude, Codex, Gemini, and more.
Download the appPlan in the dashboard
Break work into phases with titles, specs, and status. Mark them to-do or in-progress — the app picks up exactly what needs to run.
Execute with one click
Hit Execute Plan and your linked AI agent starts working through every pending phase in a live terminal — no manual prompting needed.
Your agent, your choice
Works with Claude, Codex, Gemini, Aider, Goose, and more. Link your preferred agent in settings and switch any time.
Free forever for solo developers. Paid plans unlock team features.
For solo developers.
For small teams.
For agencies and large teams.