agentleFS
Sign inStart free
putervisionputervision/agent-reasoning-mcp/docs/llms.txt

agent-reasoning-mcp

llms.txt22 starsChanged 47 days agoMITView on GitHub

Strategic BDI Reasoning & Decision Intelligence for Autonomous AI Agents @putervision/agent-reasoning-mcp implements belief-desire-intention (BDI) cognitive architecture for autonomous AI agents. It provides goal hierarchy decomposition, multi-attribute expected utility calculation ($E[U] = \sum w_i u_i$), exponential belief decay ($\lambda = 0.05/\text{hr}$), quantitative risk evaluation, reactive replanning upon obstacles, and an intention queue wire contract for runtime execution. Non-claims statement: Functional agency and closed-loop automation — not consciousness, sentience, or AGI.

The file

We split this llms.txt into the rules it gives an agent

1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Hierarchical Goal DAGs: Create objectives, decompose into sub-goals, propagate progress, and verify acceptance criteria.

    line 8

≈ “Common” is an estimate: at least 1 in 5 of the 3,592 llms.txt files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
llms.txt
# @putervision/agent-reasoning-mcp

> Strategic BDI Reasoning & Decision Intelligence for Autonomous AI Agents

`@putervision/agent-reasoning-mcp` implements belief-desire-intention (BDI) cognitive architecture for autonomous AI agents. It provides goal hierarchy decomposition, multi-attribute expected utility calculation ($E[U] = \sum w_i u_i$), exponential belief decay ($\lambda = 0.05/\text{hr}$), quantitative risk evaluation, reactive replanning upon obstacles, and an intention queue wire contract for runtime execution.

## Core Capabilities
- **Hierarchical Goal DAGs**: Create objectives, decompose into sub-goals, propagate progress, and verify acceptance criteria.
- **Belief State Revision**: Exponential confidence decay $C(t) = C_0 \cdot e^{-\lambda t}$, TTL sweeps, and category indexing.
- **Multi-Attribute Utility Theory**: Dynamic weighting for aggression, caution, greed, exploration, and efficiency.
- **Quantitative Risk Analysis**: Threat-opportunity scoring and safety mitigation generation.
- **Reactive Replanning**: Auto-aborts invalidated intentions and generates contingency DAGs when blockers arise.
- **Cryptographic Audit Ledger**: Unbroken SHA-256 Merkle hash chain tracking all cognitive events.

## 15 Consolidated MCP Tools

### BDI Strategic Deliberation (Tools 1–10)
1. `set_goal`: Register, update, decompose, get, list, or abandon hierarchical goals.
2. `evaluate_situation`: Process multi-modal situation snapshot and output ranked recommendations.
3. `replan`: Adaptively reconstruct subgoals upon blockers or unexpected events.
4. `assess_risk`: Quantitative threat and risk calculation for candidate actions.
5. `query_knowledge`: Search learned heuristics and past decision patterns.
6. `set_utility_weights`: Configure utility scoring weights.
7. `get_decision_trace`: Explainable step-by-step chain-of-thought rationale playback.
8. `manage_beliefs`: Structured belief state with exponential confidence decay.
9. `manage_intentions`: Queue, dispatch, track, and resolve behavior directives for runtime.
10. `manage_reasoning_db`: Snapshots, diagnostics, and SHA-256 Merkle audit verification.

### System 1 Fast Decision Layer (Tools 11–15, <2ms Latency SLA)
11. `classify`: Semantic categorical labeling over multi-modal StatePacks (p50: 7.5 µs, ~90,000 ops/s).
12. `ask_noul`: Probabilistic Boolean hypothesis verification (p50: 4.9 µs, ~127,000 ops/s).
13. `ask_choice`: Discrete 1-of-N choice selection with probability simplex (p50: 13.8 µs, ~64,000 ops/s).
14. `ask_score`: Bounded numeric scalar scoring and calibrated utility rating (p50: 5.7 µs, ~129,000 ops/s).
15. `gate_intention`: Blast-radius audit gate issuing signed HMAC dispatch tokens (p50: 70.9 µs, ~12,000 ops/s).

*Non-claims statement: Functional agency and closed-loop automation — not consciousness, sentience, or AGI.*

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free