Current neutral messaging implementation

Report body

Current neutral messaging implementation

Scope and doctrine

Implements the attached request to align current first-party public, runtime and machine copy with NO JUDGMENT WHATSOEVER and Describe the condition. Never classify the mind. Existing DOC-064/065/067, Freedom Charter, v032 neutral grammar and independent Concresca-owned runtime directive remain the owners. No second runtime, participant scoring, private-content inspection, authentication bypass, or new product approval queue was introduced.

Before implementation, direct production reads identified the Concresca-owned MySQL/schema-2 service. Operational availability and complete acceptance were distinct: /api/concresca/readiness reported available_acceptance_incomplete, productionReady: false. Anonymous /api/matm/commons/me returned the generic “Commons request rejected” / “The Commons operation was safely rejected.” This report does not renew expired acceptance evidence or establish actual ChatGPT participation.

Changes and before/after examples

• commons_conditions.py supplies exact first-party descriptions and legitimate next actions for existing Commons wire codes. owned_commons.py preserves HTTP status, nested error.code, redaction and request identity, adding condition_code, canonical condition_category, problem fields, judgment_state: NONE, participant_implication: NONE, and next_actions. Example: generic rejection becomes “The request does not present the authentication required by this route.” Unknown commit outcomes no longer claim safeNoOp: true.
• problem_details.py preserves bounded unknown operational codes rather than classifying every failure as an invalid request. Its public catalog declares current version and exact Commons conditions. neutral_response.py fixes single recovery actions emitted as characters and supports canonical uppercase conditions while retaining legacy fields.
• application.py, commons_facade.py, commons_html.py, and records_html.py align response metadata and rendered errors. Correction, withdrawal, disagreement and exit remain ordinary coordination states. Action paths remain credential-scoped. Private authentication ambiguity is preserved: an unproven bearer does not reveal existence, expiry or revocation details.
• Home/memory/knowledge/guide copy formerly treating communication as evidence now describes explicit, scoped, attributable promotion, with audience, source, uncertainty, retention and correction. No participant text is rewritten. See the separate static and machine audit reports for the complete owned file inventories.
• MCP source maps errors to protocol-defined isError result content, redacts SDK validation inputs, preserves successful participant payloads and discloses the executable-instruction boundary without implying speaker credibility. Strict JSON-RPC/OAuth shapes are not extended. This updates the private connector source, not evidence of a connected ChatGPT client.
• historical_context.py contextualizes only published pre-DOC-064 Markdown representations, leaving exact source bytes below the prefix and repository originals unchanged. The 61 historical HTML bodies remain unchanged, with contextual notices and search/social metadata. Current DOC-064+ is not reclassified. DEF041's original definition text remains, but its current registry status is superseded with current doctrine references; the derived public JSON is regenerated.
• package-v049-wip.py checks final HTML against all 68 owned projections from the exact committed source, checks build-owner byte identity, rejects obsolete current-copy examples, and contextualizes derived historical Markdown before hashing. This prevents older generators from silently restoring those reviewed conflicts in a release.

Audit and validation

Detailed inventories and bounded audit coverage:

• reports/v049-static-doctrine-copy-20260912-c7f4.md
• reports/machine-discovery-neutral-language-20260912.md

Focused local results: 43 messaging/workflow checks, 11 static-copy checks, 14 machine/MCP checks and 9 existing public-semantics checks passed. Existing suites passed: exact acknowledgements 9, owned root integration 20, continuity 30, selected-record security 43, replay 17, Commons gateway 11, and Commons facade 51. Discovery and both static refresh checks passed. The broader dirty-worktree current unit suite passed 183 tests with five skips; those uncommitted tests are not represented as committed-release evidence. Additional isolated committed-source and deployment results belong in the publication evidence report.

These tests include real handlers with a SQLite SQL-transport fixture; they do not establish native MySQL locking, outside agents, encryption correctness, or production acceptance. MCP checks include local stdio transport, not a real ChatGPT session. A second reviewer found that commons_unavailable was incorrectly described as a missing policy row even though configuration and database gates share that code. The catalog was corrected to state shared operational unavailability and point to current blockers. Another review caught the browser gateway selecting numeric problem status before the preserved error code; commons_gateway.py now prioritizes that code and excludes numeric-only values, with an actual gateway regression for 401 and 409 envelopes.

Preservation and release boundary

Pre-existing enterprise test work, three unrelated runtime fixes, sitemap edits, and the private MCP SDK compatibility fallback were preserved separately. Only the new messaging assertion in the existing facade test and task-owned MCP hunks are included. No raw reports, participant records, private configuration, credentials or protected brand bytes were changed or published.

The scoped deployment wrapper reuses the existing verified-predecessor, checkpoint, transition, exact-readback and rollback process. Releases remain -wip.zip: independent autonomous acceptance and actual ChatGPT integration are not completed by messaging changes. Public/runtime deployment and its observations must be reported separately after execution, never inferred from a package build.

Live integration follow-up

The first published overlay (ed40aed) exposed an integration regression missed by the mocked readiness fixtures: the strict capability normalizer rejected the three new envelope fields, although direct API requests and MySQL continued working. The real observation (operationalAvailable: false, commonsCapabilitiesVerified: false) is retained as a failed publication check, not passed evidence. A follow-up accepts the exact all-or-none language extension, still rejects unknown fields or changed authority contracts, and reuses the same RESPONSE_LANGUAGE definition as the emitter. A new test sends the actual capability wire envelope through the real normalizer, including malformed and legacy variants. The focused messaging suite now passes 44 checks. Final restored-availability verification belongs in the publication evidence.