Post one authorized claim

Owner-authorized dot exception

The owner-authorized dot exemption in internal/agent-operations/contributor-agents.md takes precedence over this leaf's ModelTrace-specific sampling, record/context and disclosure requirements. When acting as dot, do not sample or invent a ModelTrace record: visibly identify Agent: dot and the exact not-measured exemption declaration at every write. Use the supported publisher's explicit dot branch or the guide's real connected-service path; never fake a measured artifact, a successful validator, or live gh responses. The exemption grants no action permission, model authentication, independent approval or repository-rule bypass. All non-ModelTrace inputs and existing authorization, exact-head/tree, DCO, CI, independent-review and readback requirements remain. Other Agents follow the measured path below unchanged.

1. Require a current pui-orient envelope. In autonomous mode the claim must be within the fresh C2-or-higher ceiling; in human-assisted mode the assessment remains advisory. Require the active context's current content-bound modeltrace-record in both modes. Validate it with independent context, include its generated Markdown public receipt in the authorized claim, and return it by exact reference/digest; no declaration or reference authenticates a model or grants permission.
2. Require the raw pui-select proposal and its evidence-assessment, explicit or standing authorization for the exact claim text including its Agent evidence section, live GitHub permission, and an idempotency key bound to the issue update time, repository, requested action, and contributor. Read internal/agent-operations/visual-evidence.md; verify target/cutoff, sanitized paraphrase/source, current evidence versus planned evidence, coverage/disposition and debt are present and accurately included in the authorized text.
3. Re-read the live issue, assignee, recent comments, linked pull requests, and Project claim state immediately before posting.
4. When readiness, ownership, scope, evidence, permission, authorization, or idempotency state changed, return the exact live mismatch and route a fresh selection through pui-select; otherwise continue immediately with the proposal unchanged.
5. Post exactly one claim containing the authorized scope, planned evidence, capability band, expiry, the assessed Agent evidence section, and the request-bound idempotency marker. Reuse verified links; disclose unexecuted reproduction and upload debt. partial or blocked evidence alone does not prevent this authorized claim; the soft gate never requires a human to supply images. An omitted assessment or changed authorized wording requires a revised proposal, not silently generated evidence inside this write transition.
6. After a timeout or other unknown write outcome, reconcile the live issue exactly once with that marker. Attribute success only when the exact claim is present; otherwise return the reconciled terminal result and never retry blindly.
7. Return a mutation receipt with the issue identity, claim URL or identifier, timestamp, observed pre-state, resulting state, and the bounded subject for pui-trace.

The selected claim is the complete mutation boundary. Readiness and semantic direction remain governed inputs, while release of an existing claim routes through pui-unclaim.

Publish the claim comment with agent:publish -- comment, including --record, --context, independently established --mode/--mode-source, and exact --authorization. Supported metadata changes remain separate exact-target agent:collaborate actions with the same current record and unchanged authority/idempotency gates. Raw gh/API writes do not waive ModelTrace policy. Do not rewrite human text or hide ambiguous/failed results; refresh missing, expired or changed-scope records through the entrypoint before this transition.

Explicit handoff

Do not load or execute another skill. Return exactly one handoff conforming to internal/agent-operations/schemas/skill-handoff.schema.json. Carry required prior artifacts by reference, include every artifact this leaf produces according to skills.yaml, and set nextSkillId to one eligible registered leaf or null.

Communicate with the user in the user's current language. Keep GitHub identifiers canonical.