Loading ToolBench
Requirements traceability management MCP server that exposes RTMX operations as tools for AI agent integration via JSON-RPC 2.0 over HTTP and stdio
RTMX provides 10 tools with complete JSON schemas and descriptions. Naming follows verb_noun convention (status, backlog, health, deps, verify, markers, next, claim, release, release_assign). Descriptions are present and contextual (15 - 100 tokens/req noted). However, parameter descriptions lack depth, most are single-clause statements without format constraints, ranges, or examples. Output schemas are not documented. Error handling is minimal; no recovery guidance visible. Tool composition is sound (single responsibility), but descriptions could be 50 - 200 chars per pattern baseline and currently average ~60 chars. No tool annotations (readOnlyHint/destructiveHint) despite clear WRITE vs READ_ONLY risk labels in metadata.
Prioritized incomplete requirements (~27 tokens/req, filterable by category/status)
Claim a requirement for work (agent identity required, prevents concurrent work)
Dependency info: specific req ~50 tokens, overview ~19 tokens/req
Health checks on the RTM database (~100 tokens fixed)
Requirement markers in test files (~22 tokens/req, filterable by category/status)
Next requirement to work on (agent-aware, respects claims)
Release a claimed requirement (agent identity required, only owner can release)
Output schemas not documented. LLMs cannot infer response structure, forcing them to guess at field names and types for downstream tool chaining.
Parameter descriptions lack format constraints and ranges. E.g., 'limit' has no min/max; 'category' and 'status' have no enum constraints despite clear valid values (CLI, MCP, ADAPT; COMPLETE, PARTIAL, MISSING, NOT_STARTED).
No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite clear WRITE vs READ_ONLY risk labels in metadata. Agents cannot distinguish safe from destructive operations without explicit hints.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | B | 70 | 2026-07-28+ | v2 |
Release multiple requirements and assign to a version (agent identity required)
RTM completion counts and percentages (~15 tokens/req, filterable by category/status)
Run tests, map results to requirements, update database (~30 tokens/req)
Error handling and recovery guidance absent. No indication of what to do if a requirement is not found, a claim fails due to concurrent work, or a test command fails.
Parameter descriptions are minimal (single clause). Baseline for A+ tools is 72 chars per param; most here are 30 - 50 chars. Missing context on when to use optional params and what happens when omitted.