Summary
Expose the predicate graph satisfaction deficit (total unsatisfied axioms) as a single number via MCP, and integrate it into the review-plans output as a progress metric.
Motivation
The predicate graph already computes satisfaction (91.3% rate, 40 MUST errors, 417 SHOULD warnings). But this isn’t available as a summary metric or via MCP. The semiotic-project-management spec defines satisfaction deficit as THE progress measure.
Steps
- Add satisfaction_deficit tool to MCP server (returns total error + warning counts).
- Add weakest_files tool to MCP server (files with most violations).
- Integrate deficit into review-plans output.
Done when
- MCP server has satisfaction_deficit and weakest_files tools
- review-plans shows deficit
Dependencies
- Plan 0015 (MCP tooling)
Log
2026-03-07 — Created. The predicate graph satisfaction checker exists (predicate-graph.py —satisfy-all) but its output isn’t available via MCP or integrated into planning.