Loading ToolBench
Autopoietic Entity System - ISO AES-SPEC-001 Conformant. Implements Model Context Protocol for structural coupling with LLM hosts (Claude, OpenAI, Gemini). Provides read-only observation resources and state-modifying tools with guard protection.
Entity MCP server exhibits severe definition quality gaps. All 8 tools lack parameter descriptions entirely, a critical violation of pattern:tool-description. Tool names are domain-specific but lack standard verb prefixes (session_start, session_end, recharge, verify_record, op_exec, agent_cycle, coupling_grant, coupling_complete). Descriptions are present but generic and lack actionable context (e.g., 'Start a coupling session' does not explain WHEN to call it or WHAT it returns). No output schemas are documented. Input schemas exist but parameters lack type annotations and descriptions. Error handling is absent from all tool definitions. The domain (autopoietic entity systems, Lyapunov stability, event sourcing) is highly specialized, but tool definitions do not bridge the gap between domain concepts and LLM-friendly interfaces.
Force an agent sense-making cycle. The agent will feel and respond.
Complete a granted coupling request.
Grant a pending coupling request from the agent.
Execute an operation from the catalog. Requires active coupling session.
Recharge entity energy by +0.10. Use when energy is low.
End the current coupling session. Applies energy decay.
Start a coupling session with the entity. Creates a bidirectional coupling (AXM-007).
All 8 tools lack parameter descriptions. The 'partner' param in session_start, 'operation' and 'params' in op_exec, 'request_id' in coupling_grant/coupling_complete, and 'outcome' in coupling_complete have no descriptions explaining their purpose, format, or constraints.
No output schemas documented for any tool. LLMs cannot infer what fields to expect or plan downstream calls. E.g., does session_start return a session_id? Does op_exec return operation results?
Tool names lack standard action verbs. 'session_start' and 'session_end' are acceptable, but 'recharge', 'verify_record', 'op_exec', 'agent_cycle', 'coupling_grant', 'coupling_complete' are domain-specific and do not follow verb_noun convention (e.g., 'execute_operation', 'trigger_agent_cycle', 'approve_coupling_request').
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 29 | <=2025-11-25 | v2 |
Verify all invariants AND record result to event chain.
Descriptions are generic and lack context. 'Start a coupling session with the entity' does not explain prerequisites, side effects, or when to call it instead of other tools. Descriptions should be 50 - 200 chars and answer WHAT, WHEN, and WHAT IT RETURNS.
No error handling guidance. Tools declare WRITE risk but do not document what errors can occur, how to recover, or what the LLM should do next. E.g., what if session_start fails because a session is already active?