by AgentBull
The open source Agent Harness with a Company Brain. Self-hosted AI Agents for Slack, Teams, Lark, DingTalk, WeCom, and more. An open source alternative to Claude Tag.
Available after the catalog security scan passes.
⚠️ Third-Party Software Notice
This skill is third-party open-source software developed and hosted independently on GitHub. SkillsLLM lists this project for discovery and does not control or maintain the underlying repository.
Any security checks, ratings, or warnings displayed by SkillsLLM are automated and limited in scope. They do not constitute a security certification or guarantee that the software is safe, error-free, or free from malicious code, vulnerabilities, compromised dependencies, or prompt-injection risks.
Review the source code, permissions, dependencies, and configuration before installing or running any third-party skill. Use is at your own risk. To the maximum extent permitted by applicable law, SkillsLLM is not liable for losses arising from third-party software.
The deep catalog scan for this skill is still queued. Run an instant dependency check now instead.
# Add to your Claude Code skills
git clone https://github.com/AgentBull/open-ankoleGuides for using ai agents skills like open-ankole.
See how open-ankole compares with popular alternatives.
Why OpenAnkole · Company Brain · Decision work · Enterprise runtime · Architecture · Current status · Development
OpenAnkole is the open source Agent Harness in this repository. Company knowledge, live signals, permissions, tools, and observed results become the context that an Agent uses to make a decision and to show its evidence. OpenAnkole includes an enterprise bot that you can run in place of Claude Tag. You host the control plane, the workers, and the data.
Use Ankole.AI at https://ankole.ai/ for the hosted Agent OS. Ankole.AI is closed-source SaaS. In Ankole.AI, a team can change Base, Doc, and Applet while work is in progress. That is malleable software.
Give your company a Brain and every Agent better judgment.
Company Brain gives durable Agents shared knowledge. The Harness applies enterprise authority and carries work across model calls.
A model can reason only over the context it receives. OpenAnkole selects the relevant facts and capabilities, enforces authority, preserves work through failure, and records results for later decisions.
OpenAnkole runs on infrastructure that the company controls. Identity, context, credentials, artifacts, audit records, and execution stay in that environment.
The Harness gives each model call continuity, authority, durable state, and feedback from results.
Most Agent stacks combine a model, a prompt, and a set of tools. Each call begins with context assembled for that moment. OpenAnkole maintains the company state and runtime required for work that continues after the call.
Company Brain gives every authorized Agent access to the same current company knowledge.
It learns from conversations, registered files and URLs, and deliberate Agent writes. Every claim keeps its source, time, holder, confidence, and audience.
OpenAnkole supports decisions with explicit assumptions and observable outcomes. Current examples include industry research, product selection, deep data analysis, and forecasting.
OpenAnkole runs each active session as an addressable Virtual Actor. The actor can wake, receive messages, checkpoint, stream progress, hibernate, recover, and continue.
Five mechanisms keep the work alive and accountable:
Agents can work for hours or days, receive input while running, fail independently, recover with context, and record committed effects. The runtime design article explains the role of OTP in detail.
This diagram shows the boundaries for ownership and durability. Internal calls are omitted.
flowchart TB
External["External systems and operators<br/>channels · webhooks · AI API clients<br/>Console · APIs · SSO · directory"]
subgraph Control["Control Plane · one logical state and coordination boundary"]
direction TB
Platform["Principal / AuthZ / configuration<br/>Control Plane Plugins"]
SG["SignalsGateway<br/>channel ingress · webhook admission · delivery"]
Schedule["Schedule<br/>checkbacks · cron"]
Runtime["Actor Runtime<br/>session lifecycle · admission · recovery"]
Jobs["Durable work control<br/>Background Agent Jobs · Automation Jobs"]
Brain["Brain<br/>shared knowledge · recall · dreaming"]
AI["AIGateway<br/>model routing · conversations · credentials"]
end
Fabric["RuntimeFabric<br/>transient actor traffic · bounded RPC · worker files"]
Workers["Agent Computer Worker pool · 1…N<br/>Main Agent turns · Background Job / Codex · Automation scripts<br/>tools · Skills · MCP · browser · terminal"]
Providers["AI providers<br/>LLM · embedding · rerank · image · web"]
PG[("PostgreSQL · durability boundary<br/>committed domain facts")]
Home[("Shared Agent Home · durability boundary<br/>workspaces · artifacts · resumable files")]
External -->|"inputs and administration"| Control
SG -->|"ActorEvent"| Runtime
Schedule -->|"ActorEvent"| Runtime
SG -->|"bound webhook"| Jobs
Schedule -->|"bound trigger"| Jobs
Platform --> Runtime
Control -->|"live execution"| Fabric
Fabric <--> Workers
Workers -->|"AIGateway API"| Control
Control -->|"provider calls through AIGateway"| Providers
Control -.-> PG
Workers -.-> Home
The Elixir/OTP control plane owns durable decisions for Principal/AuthZ, SignalsGateway, Schedule, Actor Runtime, Job lifecycles, Brain, and AIGateway. PostgreSQL stores their domain facts.
OpenAnkole runs in production as a complete enterprise Agent Harness. Companies can host the control plane, Agent Computer, kernel, and operator console on their own infrastructure.
OpenAnkole is still defining its public API compatibility contract. Releases can include breaking changes.
| Area | Status |
|---|---|
| Control plane | Phoenix/OTP application under app/control_plane. Owns durable state, configuration, actor orchestration, Principal/AuthZ, AIGateway, Brain, SignalsGateway, and operator APIs. |
| Agent Computer | Bun/TypeScript Worker runtime under app/agent_computer. Runs the Agent loop and local tools inside an isolated Linux Worker image. Its supported role is Worker execution. |
| Kernel | Rust crate under app/kernel, loaded by Elixir (Rustler) and Bun (N-API) for crypto, identifiers, AuthZ evaluation, and RuntimeFabric envelope validation. |
| Frontend | Vite + React console, auth, and setup surfaces under app/webapps, built into the Ph |
open-ankole is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by AgentBull. The open source Agent Harness with a Company Brain. Self-hosted AI Agents for Slack, Teams, Lark, DingTalk, WeCom, and more. An open source alternative to Claude Tag. It has 217 GitHub stars.
open-ankole's catalog security scan is still queued. You can run an instant dependency and prompt-injection check now with the "Scan for vulnerabilities" button above.
Clone the repository with "git clone https://github.com/AgentBull/open-ankole" and add it to your Claude Code skills directory (see the Installation section above).
open-ankole is primarily written in Elixir. It is open-source under AgentBull on GitHub, so you can review or fork the full source.
Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh open-ankole against similar tools.
No comments yet. Be the first to share your thoughts!