MCP server for DevFlow — exposes graph queries to AI assistants. Provides tools for querying development activity graphs, DORA metrics, PR analysis, and deployment tracking.
DevFlow MCP exposes 7 tools with basic schemas and descriptions, but lacks depth in parameter documentation and output structure guidance. All tools are READ_ONLY (good for safety), but descriptions are terse (avg ~50 chars, well below the 194-char baseline). Parameter descriptions exist but are minimal. No output schemas documented. Tool names follow verb_noun convention (good), but several tools have overlapping semantics (find_blockers vs list_open_prs both show 'open PRs sorted by stuck time'). No error handling guidance, no pagination support despite tools likely returning lists. Schema completeness is moderate, types present but missing constraints, ranges, and format specifications.
Show DORA metrics (Deployment Frequency, Lead Time, Change Failure Rate, MTTR).
Explain a PR: blockers, CI, reviews, timeline.
Show all open PRs sorted by stuck time.
Show all open PRs sorted by stuck time.
Natural language graph query.
Generate standup notes for a developer.
Show event timeline for a PR.
Duplicate tool semantics: devflow_find_blockers and devflow_list_open_prs both return 'all open PRs sorted by stuck time'. LLMs will waste reasoning cycles choosing between them.
No output schemas documented. LLMs cannot plan downstream tool calls or extract required fields (e.g., PR IDs for chaining to devflow_explain_pr). Violates response-shaper pattern.
Parameter descriptions are minimal (avg ~20 chars). 'Repository identifier (optional)' lacks guidance on format (GitHub org/repo? URL? slug?). Violates tool-description pattern baseline of 72 chars per param.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 40 | <=2025-11-25 | v2 |
No pagination support visible. Tools returning lists (find_blockers, list_open_prs, dora_metrics) lack limit/offset/cursor parameters. Large result sets will blow context windows.
No error handling guidance. Tools provide no recovery hints (e.g., 'PR not found, try devflow_list_open_prs to find valid PR numbers'). Violates recovery-guide pattern.