<!-- loads: {"always":["devrites-lib/reference/standards/core.md","devrites-lib/reference/standards/code-navigation.md","devrites-lib/reference/standards/code-index.md","devrites-lib/reference/candidate-integrity.md","devrites-lib/reference/orchestration-profiles.md","devrites-lib/reference/parallel-dispatch.md","devrites-lib/reference/standards/verification-methods.md","rite-seal/reference/phase-contract.md","rite-seal/reference/final-evidence.md","rite-seal/reference/go-no-go.md","rite-seal/reference/risk-and-rollback.md","rite-seal/reference/seal-template.md","rite-seal/reference/output.md","rite-seal/reference/anti-patterns.md"],"triggers":{"agents":["devrites-lib/reference/standards/agents.md"],"browser":["devrites-lib/reference/standards/browser-proof-checklist.md"],"code-review":["devrites-lib/reference/standards/code-review.md"],"deprecation":["devrites-lib/reference/standards/deprecation.md"],"documentation":["devrites-lib/reference/standards/documentation.md"],"dod":["devrites-lib/reference/standards/definition-of-done.md"],"observability":["devrites-lib/reference/standards/observability.md"],"principles":["devrites-lib/reference/standards/principles.md"],"security":["devrites-lib/reference/standards/security.md"],"testing":["devrites-lib/reference/standards/testing.md"]},"workspace":["brief.md","spec.md","state.md","decisions.md","assumptions.md","questions.md","decision-coverage.md","architecture.md","plan.md","tasks.md","traceability.md","eng-review.md","test-plan.md","gates.md","evidence.md","touched-files.md","review.md","seal.md"],"workspaceByRole":{"code-reviewer":["spec.md","plan.md","tasks.md","test-plan.md","architecture.md","decisions.md","state.md","evidence.md","touched-files.md","review.md"],"devex-reviewer":["spec.md","plan.md","tasks.md","state.md","touched-files.md","evidence.md","devex.md","review.md"],"doubt-reviewer":["spec.md","plan.md","tasks.md","state.md","decisions.md","review.md"],"frontend-reviewer":["spec.md","plan.md","tasks.md","state.md","touched-files.md","evidence.md","review.md"],"performance-reviewer":["spec.md","plan.md","tasks.md","state.md","touched-files.md","evidence.md","review.md"],"proof-runner":["spec.md","test-plan.md","tasks.md","gates.md","evidence.md","state.md"],"security-auditor":["spec.md","plan.md","tasks.md","state.md","touched-files.md","evidence.md","review.md"],"simplifier-reviewer":["spec.md","plan.md","tasks.md","state.md","touched-files.md","review.md"],"spec-reviewer":["brief.md","spec.md","decision-coverage.md","questions.md","decisions.md","assumptions.md","state.md","review.md"],"test-analyst":["spec.md","tasks.md","test-plan.md","state.md","evidence.md","review.md"]},"rootOnly":["devrites-lib/reference/standards/code-index.md","devrites-lib/reference/orchestration-profiles.md","devrites-lib/reference/parallel-dispatch.md"]} -->

Read-set manifest: devrites-engine context <slug> --phase seal bundles every file named below into one deduplicated read. Trigger names map to the conditional rules in the sections that follow.

/rite-seal: GO / NO-GO

Verify final candidate in seal.md; /rite-ship owns authorized irreversible actions and close-out.

Authority

Read core.md plus only final-diff-relevant standards: agents, code-review, testing, browser-proof-checklist, security, principles, documentation, observability, deprecation, definition-of-done, and verification-methods.md for coverage-denominator and evidence-status judgment. Load final-evidence.md and go-no-go.md for proof and verdict safety. Follow the shared candidate-integrity.md for the reviewed digest and artifact bindings.

The host runs/reconciles named read-only reviewers; engine checks invariants. Standard is default; Full broadens conditional axes per devrites-lib/reference/orchestration-profiles.md.

Gate

State	Verdict
Acceptance proven, deterministic checks pass, no unresolved drift or human gate, Critical == 0, Important == 0	GO
Same, but Important > 0	Ask Important findings remain. Proceed to seal? [y/N]; default N. Only a human answers y; unattended (AFK/autocomplete) keeps N and stops for a human if Important remains
Critical > 0	NO-GO
Any acceptance criterion, resolved prohibition, or declared key link lacks proof	NO-GO
gates status is not all-met — an unmet, stale, unapproved, or abandoned ledger gate remains	NO-GO
Evidence is stale, or native diff review plus the exact test analyst finds weakened tests	NO-GO: Critical
Any exact required reviewer account is missing or silent	NO-GO
Any deferred finding (touched-files.md ## Review trail, eng-review.md ## Deferred findings) lacks a review.md verdict	NO-GO
Any gate: validating question remains open	NO-GO
Material UI acceptance lacks or fails required browser proof	NO-GO
Unresolved drift, unsafe migration/removal, or unapproved principle violation remains	NO-GO
check regression reports an unblessed loss vs the recorded progress baseline	NO-GO
check windows reports a deferral marker the change introduced without a windows.md waiver	NO-GO

A judgment-only criterion needs the human eye; AFK records a validating question (slice: seal) and stops. Autocomplete never auto-picks it; only a human answers.

Workflow

Follow reference/phase-contract.md (risk-and-rollback.md, seal-template.md, anti-patterns.md):

1. read workspace and final diff;
2. recheck the canonical candidate/bindings and approved repository/CI proof;
3. admit every valid same-candidate Review account; dispatch remaining required exact roles under ../devrites-lib/reference/parallel-dispatch.md in one wave, proof-runner included, on the same frozen candidate, code review through its § Engineering lanes;
4. reconcile their read-only verdicts against source and immutable evidence;
5. draft GO/NO-GO seal.md with acceptance, tests, decisions, and seven roster accounts plus proof-runner and doubt-reviewer (each engineering lane its own account);
6. write exactly one candidate binding, then run devrites-engine check seal <slug> for structure and identity, never semantics, and, when that check reports a stale readiness-input binding, run devrites-engine check drift <slug> before returning to Vet so the repair starts at the named changed, missing, or added input, and devrites-engine check regression <slug> to confirm no recorded progress fact was lost since the last checkpoint — BLOCKED is NO-GO; unproven (no baseline) does not block — plus two separate runs of devrites-engine check windows <slug>, one with --base <release-base> and one with --worktree (the only mode that scans untracked files; never pass both flags in one command, the last one wins) so each deferral marker the feature diff introduced carries a windows.md waiver row (windows.md); unwaived hits are NO-GO. The engine already refuses a gates.md ledger that is malformed, unmet, stale, unapproved, or abandoned — a seal-side gates status reading anything but all-met means Prove's evidence is not current; return there, never attest past it. Before the Seal wave, run devrites-engine gates reverify <slug> --transcripts on the frozen candidate, so every gate has a transcript bound to this digest for the proof runner. It likewise refuses a malformed notes.md or any anchor graded other than exact: devrites-engine note check <slug> --repair re-anchors moved notes; stale/ambiguous/lost ones are corrected or removed before GO. Repeat each note: unexecuted driver line check seal prints in the GO/NO-GO reply.

Corrections return through affected Prove/Review before restarting Seal under evidence validity.

Only repository scripts/CI authorize commands; never guess.

On GO

Set state.md Next step: /rite-ship and stop. A GO is a verdict, not authorization to perform an irreversible action.

Output

Use reference/output.md.