[wheakerd]

Scope

Refs #217. The current routing corpus evals/routing-v6/current.jsonl had one case each for clarify-intent, delegate-simple-task, task-planning, agents-architect, optimize-agent-usage, review-axiom-task, confirm-external-action, traceable-git-submit, and reversible-system-change, and the fixed per-route workload under evals/routing/ has no case at all for five routes. This change adds thirty-four schema v6 cases to the current corpus, outside both benchmark profiles, so that every bundled route has a positive, a near-miss no-route control, and a cross-route control, with Chinese positives for task-planning and local-web-search. Case IDs are new; no existing case changed, so the contract version stays 6 and the records bound to the v6 profiles stay verifiable. Benchmark membership for the new cases and the changed expectations for the routes the next installed-runtime release narrows belong to that release's successor contract.

The evaluation guide records the addition and the decision that the per-route workload stays frozen as the routing-context budget's fixed workload, with its retired optimize-codex-usage name mapped at validation time. The current-corpus tests take the new totals and no longer pin the number of verify-completion cases. Repository policy revision 115 records the change; the plugin version and runtime contract digest are unchanged.

Affected files and surfaces

• Shared files: None
• Codex-specific files: None
• Claude Code-specific files: None
• Documentation or validation files: evals/routing-v6/current.jsonl (34 added cases); evals/README.md; tests/test_current_routing.py, tests/test_routing_evals.py, tests/test_validator_isolation.py (totals); CHANGELOG.md, evidence/repository-policy-revisions-v1.json, evidence/runtime-identity.json, evidence/release-status.json, docs/runtime-identity.md (revision 115)

Routing and authorization impact

• Route-selection impact: None. Evaluation cases describe expected selection; they change no skill text.
• Action-authorization impact: None. Every case keeps mutationAuthorized false.
• New or changed stop conditions: None.

Documentation

evals/README.md (Current Contracts and Contract changes) describes the added coverage and the frozen fixed workload. Installation and hook commands are unchanged.

Validation

Command	Result
python3 -B scripts/check-publication.py	Passed on this exact tree: 174 black-box routing cases, 66 fixed host benchmark cases, 153 required files, 30 immutable external action and image pins
python3 -B -m unittest discover -s tests -p 'test_*.py'	OK in a disposable clone of this tree: 311 tests
git diff --check	Passed
Additional targeted checks	The enforced offline route contract agrees with every new clarify-intent case; the keyword model returns no route for eleven other new cases whose expectations rest on the descriptions, which the static check does not enforce. Per-route coverage of the current corpus after the change: every route has at least three cases, at least one near-miss control that forbids it, and at least one cross-route control. ASCII scan of the changed Markdown, JSON, and Python files: passed.

Cross-platform parity

• Shared skill changes were reviewed against Codex and Claude Code, or this change does not affect shared skills.
• Manifest, marketplace, and hook differences remain intentional and documented.
• Optional host-native validation results include the host/tool version, or are marked unavailable.

Unrelated work

• I reviewed the final diff and did not reset, hide, stage, or rewrite unrelated work.
• I did not add generated caches, disposable validation copies, local maintenance notes, or tool output.
• I called out every affected file and any routing, authorization, documentation, or parity impact above.