The CLI-first AI agent runtime.
Pipe it, schedule it, embed it. One binary, any model, MCP-native — built for autonomous work, not just chat.

Documentation · Tap Registry · Website

Octomind is an open-source AI agent client: the model calls MCP tools to do real work — read and write files, run shells, search code, delegate to sub-agents. Most agent CLIs are chat interfaces that happen to run in a terminal. Octomind is a runtime that happens to have one: the same session runs interactively, piped through stdin, as a background daemon, over WebSocket, or as an ACP sub-agent inside another agent's stack. Models, tools, roles, guardrails, budgets — all of it is TOML, no framework code.

# Interactive
octomind run developer:general

# Piped — CI, scripts, automation
echo "Explain the auth module" | octomind run developer:general --format plain

# Daemon — long-running, inject messages from anywhere
echo "watch the build" | octomind run --name watcher --daemon --format jsonl
octomind send --name watcher "run the test suite"

• Quick Start
• Benchmarks — Real PRs, Held-Out Tests
• Why Octomind?
• One Binary, Five Surfaces
• Guardrails — Policy as Code
• Cost as a Control Plane
• Sessions That Stay Sharp at Hour 4
• Intent-Driven Context
• Specialists & Taps
• Built-in MCP Tools
• Power Users — Roles, Workflows, Layers
• Installation
• Configuration
• Architecture
• Contributing
• Documentation
• License

# Install (macOS & Linux) — single Rust binary, no runtime dependencies
curl -fsSL https://raw.githubusercontent.com/muvon/octomind/master/install.sh | bash

# Sign in — models included, no API keys to manage
octomind login

# Start with a specialist — no setup required
octomind run developer:general

        Octomind v0.40.2
        Role: developer · Model: octohub:auto
        ~/your/project
> _

You're in a session with an agent that can read your code, run commands, edit files, and grow capabilities as needed. Plain-line interface with markdown rendering and shell completions — no TUI to learn, works over SSH, in tmux, in CI logs.

octomind login connects you to Octomind Cloud — a subscription that includes model access through the octohub gateway, so there's nothing to configure. Prefer your own keys? Skip login entirely and bring any provider: OpenRouter, Anthropic, OpenAI, DeepSeek, Ollama, and more. Cloud is the baseline; BYOK is always a first-class path.

developer:general (and lawyer:sg, doctor:blood, …) come from the built-in default tap muvon/tap, not your local config. The config's own default tag is assistant:concierge, so plain octomind run starts that. The banner above is illustrative (the real one renders a pixel icon to the left of the text block).

Other installs: cargo install octomind (Rust 1.95+) or build from source.

We benchmark on octobench: 25 tasks harvested from pull requests merged in 2026 — mostly after model training cutoffs — across python, php, rust, c++, and js. Each agent works in the pre-fix repo; the merged fix's own maintainer-written tests (which the agent never sees) decide pass or fail. Four agents, stock single-agent invocations, no tuning:

• The harness is the leverage. opencode ran the same model on the same endpoint at the same prices — a pure harness A/B. octomind solved 24 vs 19, at half the cost: context discipline plus supervision on exactly the deep-root-cause tasks where unsupervised agents declare victory too early.
• Worst-case pricing, still ahead. glm-5.2 ran without prompt caching (every token at list price) while Opus billed ~97% of context re-reads at 1/10 cache rates — and octomind still led on solves, cost, and wall time.
• Reproducible. Full per-case table, run artifacts, and reproduction guide: BENCHMARK.md @ 8aa3968. The story behind the benchmark: blog post.

Agent CLIs in 2026 all do interactive chat well. The gaps open up the moment you leave the keyboard:

• Autonomy needs policy, not popups. Other CLIs make the human the safety layer — every dangerous call waits for an approval click. That breaks in CI, daemons, and scheduled runs. Octomind enforces deterministic guardrails from a TOML file instead. See Guardrails.
• Config wars. No central registry. Skills here, MCP servers there, agent configs nowhere. The community calls it "solidarity in frustration." Octomind ships a tap registry: each agent is one TOML file in a Git repo, octomind run category:variant and it works.
• Generic AI hallucinates in expert domains. ChatGPT writes wrong drug dosages. Lawyers cite cases that don't exist. Multi-agent specialization is now the default architecture for serious work — Octomind runs packaged specialists, one command each.
• One generic assistant for every task. Rust debugging, blood-test interpretation, contract review — same prompt, same tools. Drift compounds.
• Sessions break at hour 4. Naive truncation drops the decisions you need. Quality collapses. You restart. Octomind's compaction is cache-aware and structurally preserving. See Sessions.
• Bills surprise you. Cursor users posting $7K daily overages. No per-task budget, no kill switch. Octomind enforces hard spending caps — before the bill, not after. See Cost.
• Context is preloaded and bloated. Most harnesses load every tool and skill up front. Octomind activates skills and capabilities on intent. See Intent-Driven Context.

The same session engine, exposed however your workflow needs it:

# ACP — drop into any multi-agent system as a sub-agent
octomind acp developer:general

# Non-interactive — the message is read from stdin (pipe it in), output as plain text
echo "Explain the auth module" | octomind run developer:general --format plain

# Structured JSONL output for pipelines
echo "List TODO items" | octomind run developer:general --format jsonl

# Daemon — keep a session alive and inject messages into it from anywhere
echo "first task" | octomind run --name watcher --daemon --format jsonl
octomind send --name watcher "now run the test suite"

# Structured output — constrain replies to a JSON Schema (structured-output models only)
echo "List TODO items as JSON" | octomind run developer:general --format jsonl --schema todos.schema.json

octomind run has no message argument — when --format is set, input comes from piped stdin. --format is run-only (it accepts only plain or jsonl; server and acp do not take it). Setting --format triggers non-interactive mode only when stdin is not a terminal: octomind run developer:general --format plain typed at a TTY stays interactive, while piping into it runs once and exits. Pass --schema <file.json> to force replies to match a JSON Schema (requires a structured-output-capable model — Anthropic models error out); see the CLI reference.

See WebSocket Server, ACP Protocol, and Daemon & Hooks for the integration modes.

One binary. Every workflow.

Other agent CLIs make the human the safety layer: every dangerous tool call pops a modal, every file write waits for a click. That works for one developer at the keyboard. It breaks the moment you point an agent at a long-running task, a CI job, or an autonomous loop.

Octomind takes the opposite position. Policy lives in scripts, not prompts. Drop a .agents/guardrails.toml in your repo and the runtime enforces it deterministically — pre-call, post-result, post-turn.

# Pre-call deny — block a class of calls before they execute
[[guard]]
match   = "shell(command=^rm\\s+-rf?)"
message = "rm -rf blocked."

# Conditional rule — only fires after the agent ran git status this session
[[guard]]
match   = "shell(command=git push)"
when    = ["+shell(command=git status)"]
message = "Review changes before pushing."

# Post-result hook — non-zero exit injects feedback into the agent's inbox
[[hook]]
match  = "text_editor(path=src/.*\\.rs)"
on     = "success"
script = ".agents/check-clippy.sh"

# Post-turn validator — fires only over the call slice since it last ran
[[validator]]
name   = "tests-pass"
roles  = ["developer"]
script = ".agents/run-tests.sh"

• Guards — pre-call deny rules. Match by capability(arg_name=regex), gate by history (+used / -unused), require loaded capabilities (has = [...]). The agent never even attempts a blocked call.
• Hooks — post-result scripts. Run after each tool returns. Non-zero exit injects stdout into the agent's inbox as a user message — clippy errors, lint failures, format diffs become automatic corrections without restarting the turn.
• Validators — post-turn scripts. Fire only over the new call-log slice (cursor-based, never re-fires on old activity). Filter by role. Output is wrapped in <validation> blocks the agent reads on its next turn. This is what replaces "approve this change?" prompts in autonomous loops.

The DSL is richer than competitor lifecycle hooks: capability+arg-regex+history+role+result-regex in one declarative file. No code to compile, no plugin to install. Designed for full automation: fits CI, daemons, scheduled runs, ACP sub-agents. Full reference: Guardrails.

The world is going autonomous. The choice isn't "ask vs auto" — it's "auto with deterministic policy" vs "auto with hope." Octomind ships the former.

Pick the right model for each step. A cheap one for routine research, a frontier one for review — per-role, per-step, mid-session swap. Real-time cost tracking and hard spending caps come for free.

# Per-role model selection — pay Opus only where it's worth it
[[roles]]
name = "researcher"
model = "openrouter:google/gemini-2.5-flash"   # cheap broad context

[[roles]]
name = "reviewer"
model = "anthropic:claude-opus-4-7"            # precision where it counts

# Hard spending limits — enforced, not advisory
max_request_spending_threshold = 0.50    # USD per request
max_session_spending_threshold = 5.00    # USD per session

• Per-role and per-layer model selection across many providers — OpenRouter, OpenAI, Anthropic, Google, DeepSeek, Amazon Bedrock, Cloudflare, and more — via octolib. Different roles can run on different vendors; new providers added there become available in Octomind automatically. See Providers & Models for the current list and supported models.
• Mid-session model swap with /model anthropic:claude-haiku-4-5. Mix providers across roles — cheap model for research, best model for execution. Cost tracked separately per provider.
• Real-time cost tracking per request and per session.
• Cache-aware token accounting (cache_read_tokens, cache_write_tokens separated from input/output).
• Hard spending thresholds with enforcement. On a session cap Octomind prompts you to continue (interactive) or auto-stops (non-interactive); on a request cap it stops execution outright — before the bill, not after.

Cursor users get $7,000 surprise bills. Octomind agents trip a budget and stop — interactive sessions ask before continuing, non-interactive runs halt automatically.

Every coding agent degrades after a few hours. Context fills. Decisions get truncated. The agent forgets why it started.

Octomind's adaptive compaction engine runs automatically:

• Cache-aware — calculates if compaction is worth it before paying for it. Never breaks the prompt-cache hit by accident.
• Pressure-tiered — compacts more aggressively as context grows.
• Structurally preserving — keeps decisions, file references, errors, dependencies; drops noise.
• Plan-aware and free-form-aware — works whether you use the plan tool or have a free-form chat.
• Fully automatic — you never think about it.

The second-order benefit: smaller context means fewer tokens, faster responses, lower cost every turn after compaction fires.

Sessions also persist: octomind run --name my-feature saves as you go, octomind run --resume my-feature (or --resume-recent) picks up where you left off — including multi-day tasks. Details: Compression, Sessions.

Work on a hard problem for 4 hours. The agent still knows what it decided in hour one.

Most agent harnesses pre-load every available tool, every skill, every instruction pack into context up front. The model sees fifty tool definitions and a wall of system prompts before you type a single character. Token bills follow. Cache misses follow. Confused tool selection follows.

Octomind inverts this. Skills and capabilities sit dormant until your intent matches them — then they activate, inject their content, and stay only as long as they're relevant. Context is a function of what you're actually trying to do.

• Meaning, not keywords. A dedicated embedding model — trained on activation traffic — scores how well your request matches each skill's description. "Help me refactor this auth flow" and "the login is broken" both find the same skill; "what's the weather" finds none.
• Hand-authored rules where precision matters. Skill authors can pin activation to file names, file contents, or exact phrases when they know better than a similarity score.
• Abstain on near-ties. When two skills score close, neither fires. Better to load nothing than the wrong thing.
• Calibrated to skip, not guess. Wrong activations bloat context and waste tokens. The system defaults to silence when in doubt.

Other agent CLIs:                       Octomind:
─────────────────────                   ──────────
1. User starts session                  1. User starts session
2. Load 50 tools into context           2. Load 5 core tools into context
3. Load 30 skills into system prompt    3. Skills sit dormant
4. User types one sentence              4. User types one sentence
5. Model picks tool from a wall         5. Embed model scores → 1 skill matches
                                           → skill content injected
                                           → 6 tools in context, not 80
                                        6. Skill goes silent again when no longer relevant

Smaller context = faster first token, lower cost per turn, fewer wrong tool calls.

It compounds with the rest:

• mcp mid-session. Even when a skill activates, the underlying MCP server only spins up if the skill actually calls it. Inactive servers = zero token cost.
• Compression interplay. A deactivated skill is dropped during compaction — its content is recoverable on next activation, not pinned forever.
• Guardrails. A guard can require has = ["filesystem-read"] and only fire when that capability is currently loaded. Policy and activation share the same capability namespace.

Details: Skills, Token Efficiency.

Most "agentic" CLIs pretend context is free. Octomind treats it as the scarcest resource in the system — and only spends it on what you actually meant.

octomind run <tag> resolves a specialist — a packaged agent with its model config, system prompt, MCP servers, and tool permissions. Not a prompt file, not a skill injection — the full stack, configured by the community, ready to run.

octomind run developer:general    # general dev, language skills auto-activate
octomind run doctor:blood         # blood-test interpretation specialist
octomind run doctor:nutrition     # nutrition specialist

What happens when you run a specialist:

→ Fetches the agent manifest from the tap registry
→ Installs required binaries automatically (skips if already present)
→ Prompts once for any credentials, saves permanently
→ Spins up the right MCP servers for this domain
→ Loads specialist model config, system prompt, tool permissions
→ Ready in ~5 seconds, not 45 minutes

Every agent has built-in power tools that let it acquire new capabilities and spawn sub-agents mid-session, without restart:

User: "Cross-reference our Postgres metrics with the deployment log"

Agent:
  → mcp.enable(postgres-mcp)        # auto-detected need, no user prompt
  → agent.spawn(log_reader)         # delegates log parsing
  → results merge mid-session
  → mcp.disable(postgres-mcp)       # cleans up
  → presents the analysis

Most agent harnesses pre-load every available tool into context. Octomind starts focused for the domain and grows only when needed. Smaller context, lower cost, faster responses, no surprise tools. See Intent-Driven Context for how activation actually works.

octomind tap yourteam/tap                 # clones github.com/yourteam/octomind-tap
octomind tap yourteam/internal ~/path     # local tap for private agents

octomind run finance:analyst              # available immediately
octomind run security:owasp

Each tap is a Git repo. Each agent is one TOML file. Pull requests are contributions.

Want to publish your expertise? A doctor:medications, a lawyer:us, a devops:terraform. One file, and everyone with that problem gets a specialist instantly. How to write a tap agent →

Octomind is MCP-native in both directions: it's a client that consumes any MCP server (stdio, Streamable HTTP, OAuth), and it exposes its own built-in servers.

Every agent has these:

view, text_editor, batch_edit, extract_lines, shell, ast_grep, list_files, workdir — file operations come from the companion octofs MCP server (ast_grep is its AST-based code search). Included by default in tap formulas that need them.

memorize, remember, forget, knowledge, relate, memory_graph — long-term memory, knowledge indexing, and relationship graphs. The companion octobrain MCP server is included by default in taps that need persistent context across sessions.

Only core, runtime, and agent are built-in MCP servers shipped in the default config. The filesystem (octofs) and brain (octobrain) servers are supplied by tap formulas — a freshly generated config won't list them.

Drop executable shebang scripts into <workdir>/.agents/tools/ and they're auto-discovered as MCP tools for that project. See Local Tools.

For most users, taps are enough. For teams and power users, the configuration system is deep — all TOML, no code.

# Per-role: independent model, temperature, MCP servers, tools, system prompt
[[roles]]
name = "senior-reviewer"
model = "anthropic:claude-opus-4-7"
temperature = 0.2
[roles.mcp]
server_refs = ["filesystem", "github"]
# view/ast_grep are octofs (filesystem) tools; create_pr comes from the github MCP server
allowed_tools = ["view", "ast_grep", "create_pr"]

# Sandbox — lock all writes to current directory
sandbox = true

# Workflows — multi-step, each step its own model and toolset
# Defined in standalone TOML files, run via CLI
octomind workflow deep_review.toml

• Roles — model, temperature, system prompt, MCP servers, tool permissions per role.
• Layers — chained AI sub-agents that run after each response.
• Guardrails — deterministic policy (guards, hooks, validators) and input pipes.
• Workflows — multi-step orchestrated task runners with validation loops.
• Supervisor — out-of-band control plane: loop/no-progress detectors, verify-gate on self-reported done, cross-session learning. See Supervisor.

See Configuration Reference for everything.

curl -fsSL https://raw.githubusercontent.com/muvon/octomind/master/install.sh | bash

Detects OS and architecture, installs to ~/.local/bin/. macOS and Linux supported. Single Rust binary, no runtime dependencies.

cargo install octomind

Requires Rust 1.95+. See Building from Source.

git clone https://github.com/muvon/octomind.git
cd octomind
cargo build --release

Option A — Octomind Cloud (recommended). One subscription, all models, zero keys:

octomind login

Device-code sign-in (like gh auth login). This mints an octohub gateway key locally — the default config already routes every request through it (model = "octohub:auto"), so you're done. Learn more: octomind.run/cloud.

Option B — bring your own keys. Octomind is fully open source and works standalone with any provider:

# OpenRouter — access to many providers with one key
export OPENROUTER_API_KEY="your_key"

# Or any specific provider
export OPENAI_API_KEY="your_key"
export ANTHROPIC_API_KEY="your_key"
export DEEPSEEK_API_KEY="your_key"

Add to ~/.bashrc or ~/.zshrc for persistence. Not signed in is a normal state, not an error — then set model = "openrouter:..." (or any provider) in your config.

octomind --version
octomind config       # generate default config
octomind run          # start your first session

Config lives at ~/.local/share/octomind/config/config.toml.

octomind config --show          # view current config
octomind config --validate      # validate config

Key areas:

• Roles — model, temperature, system prompt, MCP servers, tool permissions
• Workflows — multi-step AI processing with validation loops
• Guardrails — deterministic policy (guards, hooks, validators) and input pipes
• MCP Servers — external tools and capabilities
• Spending Limits — per-request and per-session thresholds
• Telemetry — anonymous usage stats, on by default

Full reference: Configuration Reference.

Octomind reports anonymous usage — which commands, tools and models get used, plus timings, token counts and error kinds. Never your code, prompts, file paths, tool arguments or environment values. Turn it off any of three ways:

export DO_NOT_TRACK=1           # the cross-tool standard, honoured first
export OCTOMIND_TELEMETRY=0     # per-run
# or set `telemetry = false` in config.toml

Exact field list: Telemetry.

Full list: Session Commands.

One binary. The session is the unit of work. Around it: roles (who's talking), layers and workflows (multi-step orchestration), guardrails with pipes (deterministic pre-processing and policy), adaptive compaction (long-session quality), and MCP servers (tools). All of it driven by a single resolved TOML config — no hardcoded behavior, no framework code to edit.

Embedders pick their surface: interactive CLI, ACP for multi-agent orchestration, WebSocket for IDEs and dashboards, daemon mode for long-running background agents.

See Architecture for internals.

The most impactful contribution isn't code — it's specialist agents.

Every domain expert who publishes a specialist makes Octomind useful for an entirely new audience. A cardiologist publishing doctor:medications. A tax attorney publishing lawyer:us. A security researcher publishing security:owasp. One TOML file — and everyone with that problem gets a specialist-grade AI instantly.

• How to write a tap agent
• Open issues
• Building from source
• Contributing guide

• Installation & Setup
• Quickstart
• Configuration
• Providers & Models
• Sessions
• Compression
• Roles
• MCP Tools
• Workflows
• Guardrails
• Skills
• Supervisor
• Learning
• WebSocket Server
• ACP Protocol
• CLI Reference
• Config Reference

Links above are local files in this repo; the hosted docs site mirrors them. Full index: doc/README.md.

Apache License 2.0 — see LICENSE.

Octomind by Muvon | Website | Documentation