Autonomous SDLC Governance Platform powered by IBM Bob 2.0 — implements Model Context Protocol (JSON-RPC 2.0) over stdio for AST dependency analysis, compliance gap detection, test verification, and watsonx governance payload generation
BobFlow Enterprise has three tools with basic schemas and descriptions, but significant quality gaps. All three tools have descriptions (10-100 chars), but they are vague and lack actionable context. Input schemas are present but minimal, parameters lack type constraints, enums, and detailed descriptions. No output schemas are documented. Error handling is minimal (returns {error: string} without recovery guidance). Tools lack composition clarity: scan_repository and verify_test_suite both analyze code but have overlapping concerns. No tool annotations (readOnlyHint, destructiveHint, idempotentHint). Parameter descriptions are absent or trivial (e.g., 'optional target path (unused in current implementation)' signals incomplete design). The server is STDIO-only, which caps protocol readiness at 50 and severely limits production viability.
Mermaid diagrams + markdown documentation for account service architecture and compliance requirements
AST dependency tree + compliance gap analysis
Execute local tests + return assertion statuses
No output schemas documented. LLMs cannot plan downstream tool calls or extract structured data from results. scan_repository returns a complex nested object with files[], functions[], dependencies[], complianceGaps[], but no schema is provided.
Parameter descriptions are missing or trivial. 'target' in generate_onboarding_guide is described as 'optional target path (unused in current implementation)', this signals incomplete design and confuses LLMs about when to use the parameter.
No enum constraints on format parameter. generate_onboarding_guide accepts format as a free-form string with enum values ['markdown', 'mermaid'] in the schema, but the description does not explain the constraint. LLMs may hallucinate other formats.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 43 | 2026-07-28+ | v2 |
Error handling lacks recovery guidance. toolScanRepository returns {error: 'Path not found: ...'} with no suggestion for next steps. Error responses should guide the LLM: 'Path not found. Verify the path exists or call list_directories() to explore available paths.'
Tool composition overlap. scan_repository and verify_test_suite both analyze code artifacts but serve different purposes (AST + compliance vs test execution). Naming does not clearly distinguish when to use each. Consider renaming verify_test_suite to execute_test_suite or run_tests for clarity.