How astroicers/athena works
1. Repository Role
This is the public product-facing repository for Athena, an autonomous cyber-operations platform. The production Rust OODA core (athena-core) is private; this repo exposes architecture documentation, extension guidance, a sample tool manifest, and a CI contract checker. Primary onboarding starts at README.md, with operational/API details in docs/http-api.md.
2. Autonomous OODA Control Plane
Athena’s intended core flow is Observe → Orient → Decide → Act, continuously using newly collected facts to select subsequent actions. Orient is LLM-driven with a deterministic fallback; Decide applies mandatory scope, OPSEC-budget, and Rules-of-Engagement gates before execution. This repository documents rather than implements that private control plane:
• README.md
• docs/demo-walkthrough.md
• docs/http-api.md
3. External API and Operational State Boundary
The documented stable HTTP surface covers operations, OODA phases, scheduling, facts, knowledge base, vulnerabilities, campaigns, reports, and status. This is the principal integration boundary for operators or control systems; the README explicitly distinguishes planned inbound MCP-server control from current delivered functionality.
• docs/http-api.md
• README.md
4. Extensibility and Decision Plugins
The open ecosystem separates public contracts from private decision implementations. External developers provide deterministic scoring/selection behavior through the SDK contract, enabling hot-swappable decision modules without core source access. The repository includes a Rust-based validation executable for scorer contracts.
• docs/extending.md
• ci/scorer-check/Cargo.toml
• ci/scorer-check/src/main.rs
5. MCP Tool-Container Integration
Athena orchestrates security-tool containers through MCP. Tool capabilities are declared through manifests, then reviewed, deployed through operator opt-in, discovered during Orient, and scheduled by the core. The local sample defines a tool’s container, port, ATT&CK techniques, description, and intended OODA phase.
• examples/my-tool/athena-tool.toml
• docs/extending.md
• README.md
6. Runtime, Deployment, and External Dependencies
The private v2 core is described as a modular Rust workspace; the public checker is also Rust/Cargo. Decision intelligence can use hosted, local, or mock LLM backends. Local LLM support and MCP container orchestration enable air-gapped deployment; tool containers represent the main execution-side infrastructure.
• docs/llm-config.md
• ci/scorer-check/Cargo.toml
• README.md
7. Governance and Delivery Controls
Compliance is designed as an execution-path boundary rather than post-run auditing: target scope, OPSEC noise budget, RoE policy, and human review for high-risk/suspected-honeypot actions. CI enforces published contracts and documentation quality, including scorer validation, link checking, and leak gates.
• .github/workflows/contracts.yml
• .github/scripts/leak-gate.sh
• .github/scripts/check-links.sh
Main entry points: README.md Product onboarding · contracts.yml Contracts workflow · extending.md Extension guide · http-api.md Stable HTTP API
Components 19
Public contracts
Documentation and extension artifacts published in this repository.
- Product onboarding · Entry point for platform capabilities and current boundaries.
README.md - Stable HTTP API · Operator integration surface for operations, state, campaigns, and reports.
docs/http-api.md - OODA walkthrough
docs/demo-walkthrough.md - Extension guide · Defines decision-plugin and MCP tool extension workflows.
docs/extending.md - LLM deployment options · Hosted, local, and mock backend guidance.
docs/llm-config.md - Sample tool manifest · Declares container, port, ATT&CK techniques, and OODA phase.
examples/my-tool/athena-tool.toml - Decision plugins · Hot-swappable deterministic scoring and selection modules.
docs/extending.md
Private control plane
Documented behavior implemented by the private athena-core.
- athena-core · Production autonomous cyber-operations runtime.
- Observe · Collects operational facts.
- Orient · Uses LLM reasoning with deterministic fallback.
- Decide · Applies mandatory execution gates.
- Act · Schedules approved tool execution.
- Execution governance · Scope, OPSEC budget, RoE, and human review constrain actions.
- MCP tool containers · Operator-opt-in security tools orchestrated by the core.
Delivery controls
Automated checks protecting published contracts.
- Scorer contract checker · Validates published scorer contracts.
ci/scorer-check/src/main.rs - Checker package
ci/scorer-check/Cargo.toml - Contracts workflow · Runs contract and documentation quality checks.
.github/workflows/contracts.yml - Leak gate
.github/scripts/leak-gate.sh - Link checker
.github/scripts/check-links.sh
Connections 22
- Product onboarding → Stable HTTP API: links to
- Product onboarding → OODA walkthrough: introduces
- Product onboarding → Extension guide: directs extensions to
- Stable HTTP API → athena-core: documents boundary for
- OODA walkthrough → athena-core: documents behavior of
- athena-core → Observe: runs
- Observe → Orient: facts
- Orient → Decide: candidate actions
- Decide → Execution governance: must pass
- Execution governance → Act: approves
- Act → Observe: new facts
- Decision plugins → Orient: provides scoring
- Extension guide → Decision plugins: defines
- Sample tool manifest → MCP tool containers: declares
- Extension guide → Sample tool manifest: guides authoring
- Orient → MCP tool containers: discovers capabilities
- Act → MCP tool containers: orchestrates
- LLM deployment options → Orient: configures backends
- Checker package → Scorer contract checker: builds
- Contracts workflow → Scorer contract checker: runs
- Contracts workflow → Leak gate: runs
- Contracts workflow → Link checker: runs