Enterprise SDLC Automation Platform - a multi-agent software development lifecycle pipeline with FastAPI backend and Streamlit frontend
Server exposes 6 tools via HTTP FastAPI routes. All tools have basic descriptions (10-80 chars) and input schemas with type definitions. However, descriptions are minimal and lack LLM-optimization guidance (e.g., no 'when to use' context, no recovery hints). Parameter descriptions exist but are terse. Output schemas are not documented. Error handling is absent from tool definitions. No tool annotations (readOnlyHint, destructiveHint, idempotentHint). Tool naming follows verb_noun convention (create_run, resume_run, get_run_state, list_checkpoints, rewind_run, health) but lacks clarity on state transitions and side effects.
Create and start a new SDLC pipeline run with an initial idea
Retrieve the current state and status of a running or completed SDLC pipeline run
Health check endpoint for the Agentic SDLC API
List all checkpoints for a given SDLC pipeline run thread
Resume a paused SDLC pipeline run with human-in-the-loop verdict and comment
Rewind a SDLC pipeline run to a specific checkpoint for replay or correction
Output schemas not documented. Tools return responses but LLMs cannot infer field structure, forcing guesswork on downstream tool chaining.
Descriptions lack LLM-optimization context. No 'when to use' guidance, no recovery hints, no prerequisites. E.g., 'Resume a paused SDLC pipeline run' does not explain when a run becomes paused or what verdict values are valid.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). LLMs cannot determine which tools are safe to retry, which modify state, or which are read-only without explicit hints.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 59 | <=2025-11-25 | v2 |
Parameter descriptions are terse (10-30 chars). 'The thread identifier' does not explain format, constraints, or how to obtain one. LLMs cannot validate inputs without explicit constraints.
No error handling guidance in tool definitions. Tools do not document what errors are retryable, user-fixable, or fatal, or how to recover.