MCP Verify v1.0.550 Final Semantic Cleanup

Date: 2026-08-11

This release uses v1.0.549 as its baseline and completes the public semantic cleanup without changing product architecture, scoring thresholds, policy decisions, ranking inputs, or legacy machine-readable fields.

Implemented behavior

Publisher messaging now describes evidence, ownership, revalidation, and badges without implying that payment or claiming directly improves score.
Public coverage exposes validated_servers, based on completed validation timestamps, separately from the scored/ranking population. A failed completed validation is therefore validated but not scored.
Generic freshness displays use the canonical 24-hour evidence state from the current snapshot. The separate 48-hour window remains only under the explicitly named Strong live and Recent validation (48h) concepts.
Pricing and TrustOps render the same centrally sourced commitments: Community 720h target (best effort), Pro 168h target (best effort), and Enterprise 24h SLA (guaranteed).
Pricing and TrustOps define evidence freshness, validation targets, and contractual SLAs separately. Policy-specific runtime thresholds are labeled evidence-age limits.
Server summaries render Production readiness, Production trust decision, and Recommended runtime policy as three distinct concepts.
Homepage, search, ranking, shortlist, team, maintainer, Trust Index, and compare surfaces label Production readiness explicitly.
Trust Index tables render production_readiness_code under Production readiness and label the export link Policy export. The legacy production_trust_decision API and CSV field remains available.
Compare retains a separate Production trust decision row and reports unknown when that decision is unavailable instead of falling back to readiness.

Compatibility boundaries

No plan-aware scheduler was added.
No score, readiness, policy, or ranking rule changed.
No API schema or legacy machine field was removed or renamed.
Billing and claim state remain excluded from score and ranking inputs.

Regression coverage

The release adds fixtures and assertions for:

a 36-hour-old validation that is stale under the canonical 24-hour definition but remains eligible for the separately labeled 48-hour strong-live cohort;
a failed completed validation that counts as validated but not scored;
identical pricing and TrustOps commitments and terminology;
coexistence of Production readiness: Safe for evaluation and Recommended policy: Allow with approval on server and compare output;
explicit Production readiness labels across affected public tables;
continued billing neutrality and ranking independence.

Verification

PYTHONPATH=verify/src python3 -m pytest verify/tests -q --no-cov: 527 passed.
make test-suites: 293 passed; coverage quality gates passed.
python3 scripts/export_openapi.py --check: OpenAPI is fresh after version metadata regeneration.

Deployment and cache verification

Production uses the IONOS deploy/ionos/deploy.sh runbook. It sets the release SHA and v1.0.550 metadata, rebuilds and restarts the single Verify stack, waits for health, and verifies /, /pricing, /trust-index, a server detail page, /methodology, and /status against the deployed build identity.