Try Demo
AI that thinks,
remembers, and acts
Captain Claw is your personal AI workspace. Research the web, process documents and video, automate browsers, run teams of AI agents, and build reusable Flows you drive by chat or voice — and it learns from every interaction.
Works with your favorite AI models
Sign in with ChatGPT (no API key) · bring your own key · or run 100% local with Ollama
Everything you need in one workspace
44 built-in tools, 6 memory layers, autonomous thinking, multi-agent orchestration, and composable Flows — all in a beautiful web interface.
Research Workspace
Ask anything. Captain Claw searches the web, scrapes pages, processes documents, and synthesizes polished reports with tables, charts, and insights.
It Actually Remembers You
Six layers of memory — from working context to deep long-term storage. Captain Claw recalls your preferences, past conversations, contacts, and decisions.
Autonomous Thinking
A nervous system that dreams during idle hours — discovering patterns, forming hypotheses, and connecting insights you never explicitly asked about.
Beautiful Visual Reports
Generates polished HTML reports with styling, tables, and data visualizations. Export to PDF or share directly from the workspace.
Browser Automation
60+ browser actions powered by Playwright. Record workflows once, replay with different data. Tab-aware research with nested context.
Multi-Agent Orchestration
Flight Deck command center, Agent Forge team builder, Agent Council deliberation, DAG orchestrator, and BotPort agent-to-agent network — all with real-time tracking.
Document Processing
Drop in PDFs, Word docs, Excel sheets, PowerPoint decks, images, and CSVs. Captain Claw extracts, analyzes, and transforms them into actionable insights.
Google Workspace
Connected to Drive, Docs, Sheets, Gmail, and Calendar. Search your files, read emails, manage events — all from a single conversation.
Safety by Design
Three-layer guard system checks inputs, outputs, and tool executions. Approval gates for sensitive operations. Full audit trail with traceable lineage.
Video & Voice Understanding
Drop in a clip and Captain Claw samples frames, transcribes the audio with timestamps, and synthesizes one coherent description. Voice notes in, spoken replies out.
Plan Mode
Turns a single request into a reviewable DAG of steps, runs them under the orchestrator, judges each against acceptance criteria, and auto-revises what fails.
Code Apps
Ask for a tool and the agent scaffolds a sandboxed mini-app — its own backend and UI — running behind Flight Deck. Apps share data through a typed SDK.
Flows
Verb-first applications. An intent arrives on a channel, agents do the thinking, the result comes back — composable, inspectable, and increasingly authored by the system itself.
A deterministic spine, agent judgment at the leaves
A Flow is a real program — triggers, branches, calls, parallelism, error handlers — and the open-ended thinking happens only in the steps where you want it. Every flow is a small text program, validated by a deterministic parser. The model is never the source of truth; the parser is.
Compose — flows call flows with gosub, pass arguments, and return values. Flows are functions.
Parallelize — spawn / join run work concurrently. Three lookups that took 9s in series finish in 3.
Recover — on error handlers and inline status branches. No silent half-failures.
Control mid-flight — /flow status, pause, stop from any channel, or buttons in the run log.
flow "Gift Idea"
trigger any when contains "gift idea"
step who:
input
prompt: "Who is it for, occasion + budget?"
step ideas:
agent on origin
prompt: "5 gift ideas for {{steps.who.output}}"
step reply:
emit "{{steps.ideas.output}}"
output -> same
Write the DSL, build it visually, or describe it in plain English — all round-trip losslessly.
Agents write their own flows
The synthesize_flow tool turns a repeated task into a validated, call-only flow stored in a
curated scratch space. A synthesized flow earns its way to permanence — three clean runs make it a ⭐ candidate,
three failures quarantine it. Promote the good ones into your permanent flows. Agent-written automations can never borrow your vetted
authority until you approve them.
Flight Deck
A multi-agent command center. Spawn, monitor, and coordinate teams of AI agents from a single dashboard.
Agent Desktop — monitor, chat with, and configure all agents from one place
Agent Forge
Describe a business goal in plain text. An LLM designs a specialized team with roles, tools, operating procedures, and a lead coordinator.
AI decomposes your objective into agent roles
Review and customize before spawning
Each agent gets tailored tools from 44 available
Fleet instructions with Standard Operating Procedures
Agent Council
Structured multi-agent deliberation. Run brainstorms, debates, reviews, or planning sessions with 2-N agents in moderated rounds.
Agents self-score suitability per topic
Choose actions: answer, challenge, refine, broaden
Moderator synthesizes conclusions, all agents vote
Export as markdown minutes
Fleet Communication
Agents discover peers automatically and collaborate. Synchronous consult or asynchronous delegation with shared workspace.
Auto-discovery of peer agents in the fleet
File transfer and shared clipboard across agents
Director panel with fleet-wide broadcast
Per-agent token/cost analytics and trace timelines
From goal to team in one click
Describe what you need — "scout startups in Croatia, research founders, build an investment pipeline" — and Agent Forge designs the entire team.
Describe your objective
Type a business goal in plain English
AI designs the team
Roles, tools, models, and operating procedures — all generated
Review, customize, spawn
Edit anything before deployment — then spawn the entire team
Structured multi-agent deliberation
Run brainstorms, debates, code reviews, or strategic planning sessions. Each agent brings a different perspective — a financial analyst, a tech strategist, a creative director — and they deliberate in structured rounds.
2-N
Agents per council
8
Session types
5
Agent actions
MD
Export minutes
$ captain-claw-fd
Flight Deck running at http://localhost:25080
See it in action
From question to polished report in minutes
How it works
Three steps from question to insight
Ask anything
Type your question, attach files, or describe a task. Captain Claw understands context and picks the right approach automatically.
Watch it work
See every tool call, web search, and processing step in the real-time activity log. Approve sensitive operations through safety gates.
Get results
Receive structured answers, visual reports, data tables, exported files — and explore further with suggested next steps.
An AI that grows with you
Unlike chatbots that forget everything between sessions, Captain Claw maintains a rich cognitive model across six memory layers.
Working Memory
Current conversation context with smart compaction
Semantic Memory
Vector + BM25 hybrid search across all sessions
Deep Memory
Long-term archive with full-text search
Insights
Auto-extracted facts, contacts, decisions, and deadlines
Nervous System
Autonomous dreaming that discovers non-obvious connections
Self-Reflection
Periodic self-assessment that generates improvement directives
Make it yours
14 built-in themes from Amiga Workbench to Windows 11. Switch personas to change how the AI interacts with you. Choose your model per session. Create and share custom themes.
14
Built-in themes
5+
AI personas
44
Built-in tools
BYOK
Bring your own key
Built for real work
From quick research to complex multi-step automation
Research & Analysis
"Summarize the top 10 Hacker News posts with engagement analysis" — Captain Claw fetches, processes, and delivers a polished visual report.
Document Processing
"Extract key data from these 20 invoices and create a summary spreadsheet" — Drop PDFs, get structured output.
Email & Communication
"Check my inbox for urgent items and draft replies" — Connected to Gmail, understands context from past conversations.
Competitive Intelligence
"Monitor competitor pricing pages weekly and flag changes" — Schedule recurring tasks with built-in cron and playbooks.
Data Collection & Automation
"Scrape this directory, extract contact details, and save to a structured database" — Browser automation + datastore + export.
Personal Assistant
"What did I discuss with John last week about the contract?" — Cross-session memory recall with full context awareness.
Available everywhere
Access Captain Claw from any device or platform
Web App
Desktop
Telegram
Slack
Discord
X / Twitter
Ray-Ban Glasses
Deploy your way
Self-hosted, fully private, no vendor lock-in
Docker
One-command deployment with Docker Compose. Persistent data, auto-restart, ready for production.
pip install
Install from PyPI in seconds. Python 3.11+ with interactive onboarding to configure models.
Standalone Binary
Pre-built executables for macOS, Linux, and Windows. No Python required — just download and run.
$ pip install captain-claw
$ captain-claw-web
Server running at http://localhost:23080
Try it right now
Two live demos, no sign-up required. Both running in Docker.
Computer Mode
The public research workspace. Bring your own API key, get a session code, and start exploring. Great for trying the research and report generation features.
- BYOK — use your own OpenAI, Anthropic, or Gemini key
- Per-session isolation with access codes
- Research workspace with visual report generation
Full Demo
The real deal. Full access to every feature — chat, orchestrator, brain graph, memory system, playbooks, personas, scheduling, and all 44 tools.
- All 24+ modules and dashboards
- Persistent memory, brain graph, nervous system
- Multi-agent orchestration and cron scheduling
Both demos run in Docker and are periodically reset. For a permanent setup, install Captain Claw on your own machine.
Open source. Free forever.
Captain Claw is MIT licensed. Your data stays on your machine. Bring your own API keys, or run 100% locally with Ollama.
Star on GitHub