@juhlabs/openclaw-bdi

BDI governance plugin for OpenClaw agents. Makes any OpenClaw skill smarter with intent verification, audit trails, memory, and trust scoring.

Built by JuhLabs.

Install

openclaw plugins install @juhlabs/openclaw-bdi

Or via npm:

npm install @juhlabs/openclaw-bdi

What it gives your agents

| Tool | Purpose | |------|---------| | bdi_verify_intent | Challenge risky or vague actions before execution | | bdi_recall | Retrieve past experiences and learned principles | | bdi_log_decision | SHA-256 hash-chained audit trail for every decision | | bdi_reflect | Extract principles from plan-vs-actual divergence | | bdi_assess | Estimate task difficulty and check model trust scores |

Plus automatic hooks: governance rules injected before every prompt, audit trail + trust scoring updated after every agent run.

How it works

User Intent ──→ Verify Intent ──→ Recall Memory ──→ Execute
                     │                  │               │
              risk assessment    past approaches    audit logged
                     │                  │               │
              block if risky    pre-load wisdom    trust updated
                     │                  │               │
                     └──── Reflect on outcomes ◄────────┘

• Intent Verification — Flags destructive actions, credential operations, and complex tasks before execution
• Memory Recall — Retrieves relevant past experiences with speculation from failures
• Audit Trail — SHA-256 hash-chained, append-only, tamper-detectable decision log
• Trust Scoring — Tracks agent reliability per model tier, modulates routing confidence
• Principle Extraction — Learns from plan-vs-actual divergence, applies to future tasks

Compatible with

Works with all 13,700+ OpenClaw skills. Any LLM — Claude, GPT, Gemini, DeepSeek, and more.

Stats

• 28 tests, 0 failures
• 5 tools, 2 hooks, 7 core modules
• Zero external dependencies beyond zod

License

MIT