Mission control for Claude Code.
CCC runs your fleet of real Claude Code terminals — sessions, autonomous loops, and multi-agent pipelines — from one command deck.
- 8 apps · 8 packages
- real PTY — output never intercepted
- 55 MCP tools
- SQLite, your box, your data
- free for personal use
Sessions
It's a real terminal. We just gave it a command deck.
Real PTY terminals in the browser. Input to stdin, output from stdout, nothing intercepted.
Autonomy
Loops that don't fool themselves.
Ralph
Fresh-context iteration against a PRD until every story passes.
BigRalph
A skeptical second session inspects the real working dir after each run, judges the original goal, and refines the next loop. It remembers.
Intents
Write the intent. Review the PR.
Describe the outcome. CCC researches the repo, writes the plan, executes it, and opens a PR for your sign-off.
01 · intent
❯ Rate-limit the public API, behind a config flag, with tests.
02 · research
- src/api/router.tsread ✓
- src/api/middleware/auth.tsread ✓
- src/config/flags.tsread ✓
- src/lib/redis.tsread ✓
- test/api/router.test.tsread ✓
03 · plan
- 1. Add limiter middleware (token bucket)
- 2. Wire RATE_LIMIT flag into config/flags.ts
- 3. Mount on public routes only
- 4. Cover burst + steady-state in tests
04 · execute
- ✓limiter middleware addedpasses: true
- ✓config flag wiredpasses: true
- ✓public routes coveredpasses: true
- ✓tests passpasses: true
05 · review
feat: rate-limit public API behind config flag
ccc/intent/rate-limit → main
Specula
Every session, one wall.
Every live terminal tiled on one resizable wall.
Screenshots
This is the actual product.
Real captures from the running deck — not mockups.
Sessions
A live Claude Code session on the command deck — chat and terminal side by side, with context and model state in view.
Ralph loops
Autonomous iteration loops running against real repos, each with its judge tier and outer/inner run progress.
Pipelines
Multi-agent DAG pipelines with their recent runs — stage counts, pass/fail bars, and last-run timing.
Autopilot
Put the repo on autopilot. Keep the stick.
One reviewable draft PR at a time. Never merges on its own.
Captain
Scans the repo, picks the next improvement, and opens the draft PR.
First Officer
Reviews the Captain's diff and posts findings before it reaches you.
Flight Engineer
Verify-only: runs the checks and reports back. Never opens PRs.
- #41merged by you
- #42merged by you
- #43merged by you
- #44draft
Pipelines
Wire agents together. Visually.
A visual DAG of Ralphs, sessions, gates, and imagegen nodes passing {{variables}} on a shared blackboard.
click ▶ run — or tap any node
run logstdout · live
— press ▶ Run to fly this plan —
Forge
Skills earn their place.
Challenger skills fight the champion on golden tasks. You promote. Rollback is instant.
Golden-task scoreboard: challenger@v4 passes 5 of 6 golden tasks against champion@v3's 4 of 6. Promotion stays human gated — you decide.
ChatOps
DM a bot. Get a session.
Bots
One bot per project, plus one global bot for the whole deck. Slack or Discord — same commands.
DM → session
DM the global bot and a project picker appears. Choose one, and the conversation becomes a live Claude Code session.
Channels
CCC provisions a channel per project. Start a thread and it spawns a session bound to that thread — replies land where you asked.
Identities & policies
Every chat user maps to a CCC identity with a policy tier that decides what they can touch.
unrestrictedwiki_onlychat_onlycustom
Knowledge
The memory layer.
Intelligence
Mines your session history into a ranked what-to-work-on-next, plus a yesterday recap.
LLM Wiki
A living knowledge base auto-linked to projects and sessions.
Research
Long-running deep-research agents that return cited reports.
Ask the Codebase
Local-embedding RAG with file:line citations verified at answer time.
src/pty/manager.ts:214 ✓ verified
Platform
Runs on your hardware.
Worker Nodes
Pool your GPUs. Dial-home nodes serve local models; CCC routes each job to the best fit.
Media
Images and video on your own stable-diffusion.cpp host. Bring your models — FLUX, ERNIE, Wan2.2, whatever you run.
Local inference
Flip the router from cloud to local models. Offline mode built in.
cloud → localoffline mode
Computer Use
Isolated desktop control on Linux, Windows, macOS, or Docker — never your operator console.
Surfaces
The deck follows you.
DM a bot, get a session. Per-project channels. Policies decide who can touch what.
- Slack
- Discord
- MCP
- 55 tools
- CLI
- tray
- Local Agent PWA
- Remote Hosts
- Pipelines
- Autopilot
- Ralph loops
- Triggers
- Worktrees
- Wiki
- Webhooks
- REST API
G2 glasses
voice: "approve"Fleet HUD on your face. Dictate prompts, approve permission requests, hands-free.
Governance
Autonomy you can audit.
- RBAC — Google & Microsoft OAuth
- audit log
- decision logs
- token usage ledger
- encrypted secrets
- worktree isolation
- dead-man heartbeat
- automatic CLAUDE.md management & scoped skill injection
License
Pick your license.
Free
Personal & noncommercial
Everything in CCC, free for personal use under the PolyForm Noncommercial license.
Commercial
Licensed
Running CCC at work? A paid plan includes a commercial license for as long as it's active. Need one on its own? Join the waitlist and we'll reach out.
$ join --waitlist