Rite temper

Codex compatibility

This is the Codex mirror of a DevRites skill. In Codex:

• Load DevRites engineering standards from .agents/skills/devrites-lib/reference/standards/. Read .agents/skills/devrites-lib/reference/standards/core.md before workflow work, then load the other .agents/skills/devrites-lib/reference/standards/*.md files exactly when this skill asks for them.
• Use the installed devrites-engine binary as the canonical runtime helper surface for orientation, gates, and state mutation.
• Invocation and dispatch are different: invoke means run a skill in this context; dispatch means start a fresh agent with spawn_agent, await it, and reconcile its result. Never describe inline skill work as a dispatch.
• On MultiAgent V2, call spawn_agent with the exact named agent_type=devrites-<role>, a unique task_name, and fork_turns="none". Codex loads that role TOML's developer_instructions natively. Because V2 collaboration lifecycle calls bypass hooks, DevRites verifies the current durable parent/child rollout for the exact role, wait, completion, and non-empty delivered result.
• On MultiAgent V1, when the named role is not exposed, use generic explorer for a read-only role with fork_turns="none" and name exactly one .codex/agents/devrites-<role>.toml contract in the message. Trusted .codex/hooks.json injects that contract's exact developer_instructions and binds the child to the fail-closed reviewer read-only guard.
• On MultiAgent V1, devrites-slice-wright uses generic worker with fork_turns="none" and the exact role TOML named in the message. Trusted .codex/hooks.json binds it to the active reconcile window and .wright-allowlist; do not substitute worker for an exposed V2 named role.
• The invoked skill's required-agent-roles frontmatter arms the fail-closed Stop receipt. Every listed role must have a confirmed start, wait, and non-empty result in this turn.
• If any required named or generic agent dispatch is unavailable or rejected, stop for HITL. Never execute a DevRites specialist role in the root context.
• Wait for every required fresh-context dispatch before reconciling or advancing. A backgrounded or lost result is incomplete.
• Codex project hooks are installed in .codex/hooks.json; declared-leaf hooks are scoped inside .codex/agents/devrites-*.toml. Review and trust them with /hooks before relying on hook enforcement.
• When this skill asks a HITL question via AskUserQuestion: Codex's equivalent (request_user_input) exists only in Plan mode. Outside Plan mode, render the option set as a plain numbered list in chat and end the turn so the human answers: NEVER silently pick an option yourself; auto-picking is AFK's contract, gated by the .devrites/AFK sentinel.

$rite-temper: review scope and risk before planning

Review a readied spec for outcome ambition, scope, pre-mortem risks, and unnecessary solution surface. Fold accepted decisions into the canonical contract before $rite-define. This step is optional for small work and skips low-stakes specs, but $rite-autocomplete always invokes it. Read the active workspace first; if there is no readied spec.md, tell the user to run $rite-spec.

Rules consulted (read on demand from .agents/skills/devrites-lib/reference/standards/)

Pull on demand: patterns.md + coding-style.md (the over-engineering / YAGNI rubric (reuse the pack's standard, don't invent one), documentation.md (ADR-style decisions.md entries), afk-hitl.md (irreversible-risk list + gate ceiling), elicitation.md (the move-set to deepen a section that needs more than the default pre-mortem) selected by the section's risk).

Operating rules

• Raise outcome ambition without expanding the solution unnecessarily. Solve the underlying problem without adding speculative capability or abstraction.
• Route every scope change through the Spec Drift Guard. A scope change takes effect only when spec.md records the confirmed decision. Expansion is opt-in (HITL confirm; AFK pauses: see autocomplete policy). Nothing auto-grows into the build.
• Apply maximum caution to hard-to-reverse changes. Auth, migration, public API, and data-model changes always pause under the irreversible-risk list.
• Honest verdict. Never round "needs work" up to "ready"; record every scope call's why.
• Search before asking. Apply afk-hitl.md decision ownership; only human-owned calls enter the interactive walk.
• You write; the reviewer judges. You are the single canonical writer (strategy.md + the spec edits); the reviewer agent is read-only.
• Require spec-quality checklists for significant features. Before hardening, confirm checklists/<domain>.md exist and pass (rite-spec/reference/spec-checklists.md); a scope expansion you fold in adds its own rows (new Success/Acceptance criteria get checklist rows too). A folded expansion with an unquantified criterion is a CRITICAL failure that reopens the readiness gate.

Workflow

0. Read .agents/skills/devrites-lib/reference/standards/core.md, then run devrites-engine preamble for deterministic workspace orientation. Then read the workspace: spec.md, decision-coverage.md (+ decisions.md, assumptions.md, design-brief.md if UI), state.md. Require Spec gate: passed: else STOP → $rite-spec. Require Decision coverage: CLEAR: else STOP → $rite-clarify. If a plan already exists, scope changes route through $rite-plan repair (record in drift.md), not a blind spec edit.
1. Significance test: reference/significance.md. Low-stakes / shape-not-meaning work → write the exact one-line skipped — low stakes (<trigger>) verdict to strategy.md, set state.md Next step: $rite-define, and recommend it. Otherwise fire the full pass below.
2. FORWARD pass + mode selection: reference/scope-modes.md. First, the one-sentence-intent test: state the whole change's intent in a single sentence. If you can't without an "and" that joins two unrelated outcomes, it is two features: the scope-creep signal; recommend splitting or narrowing the spec before continuing. Then consider the 10-star outcome for the underlying problem and choose exactly one scope mode (expand opt-in · selective · hold-rigor · reduce-to-MVP) with its rationale and the condition that would change the choice. $ARGUMENTS --mode is a hint, not a command.
3. INVERSION pass: pre-mortem in past tense ("it shipped and failed: what went wrong"), each top risk carrying likelihood + mitigation + the slice it will bind to; then the YAGNI ledger (each candidate scope item gets the "imagine the later refactor" test; defer unless now-cost is trivial AND deferred-cost is large). Interruption pre-mortem: audit the spec forecast and assumptions for unresolved behavior, proof prerequisites, approvals, access, and irreversible gates. Resolve facts and reversible details now; retain only unavailable-pre-code or mandatory action-time checkpoints.
   • Deepen on demand. When a scope decision, requirement, or risk needs more analysis than the default pre-mortem provides, choose 3-5 techniques from elicitation.md whose when-to-reach-for-it matches that section's risk (irreversible decision → Red-Team/Blue-Team + Assumption Audit; sizing → Delphi; vague requirement → Steelman-then-Attack), offer them, and run the chosen one on that section. Record what changed, not the technique name. The default passes remain sufficient when no deeper analysis is needed.
4. Score the 9 dimensions: reference/review-dimensions.md. Cite evidence before the band; gate on the floor (the weakest dimension, not an average). Completion: all nine dimensions have cited evidence, a band, and one explicit floor verdict.
5. Review human-owned findings with the human; do not batch them. strategy.md records the interactive review but does not replace it. Present each material scope decision through AskUserQuestion, one at a time, best-guess + why. Each material scope call ends as a recorded decision: a resolved questions.md qid (HITL) or a decisions.md ADR (AFK): so the review leaves an auditable record outside chat. (AFK gate policy is single-sourced in reference/significance.md: hold-rigor + reduce-to-MVP auto-apply, any expand is a blocking pause, irreversible-risk always pauses.) Apply objective clarity, mitigation, and assumption fixes directly.
6. Write strategy.md + fold back: reference/strategy-template.md. Choose the drift path based on whether a plan exists: no plan.md yet (the normal pre-define case) → update spec.md through the Spec Drift Guard. Success criteria and Acceptance criteria for each opt-in expansion / cut, Non-goals for every deferred item, Constraints and Risks the pre-mortem demands, and the gaps/decisions table; plan.md already exists → do not edit spec.md here: write the deltas to drift.md and hand off to $rite-plan repair. Either way, append decisions.md (one ADR per scope call: context · decision · why-not · what-would-change-it) and assumptions.md (every "we'll probably need X" → assumption-to-verify). Every scope delta must carry a recorded human decision: a resolved questions.md qid (HITL) or a decisions.md ADR (AFK within the ceiling); a folded change with no recorded decision is invalid. Re-check the spec Readiness gate (it fails if any folded scope delta lacks its decision or leaves a foreseeable human build choice). After any edit to brief.md, spec.md, decisions.md, assumptions.md, or questions.md, re-scan the affected coverage rows, assumption audit, residual uncertainty, and closed gates. Partial/Missing, an unowned material assumption, or an open blocking/escalating question routes $rite-clarify/HITL; never refresh past it. Only after the matrix is re-closed, run devrites-engine readiness-digest coverage <slug> and replace the complete Coverage inputs SHA-256 line in decision-coverage.md. Update state.md: Phase: temper, Next step: $rite-define; on a blocking pause (expand / irreversible / a dimension still below bar) write the Awaiting human block + Status: awaiting_human before stopping.
7. Adversarial verification loop: dispatch devrites-strategy-reviewer through the file-backed fresh-context contract in agents.md, with only the hardened spec + rubric: no authoring reasoning. Resolve actionable findings by editing spec.md/strategy.md, re-dispatch; cap ≤3 iterations. A dimension still below bar after 3 is classified by decision ownership: a product/scope/risk choice becomes a blocking question; an objective spec defect stays blocked with the exact required edit and routes to $rite-spec, not $rite-resolve. Irreversible-risk findings always pause. Use the shared capability ladder; if no fresh-agent rung is available, stop for HITL. After an accepted edit to a coverage-bound input, repeat step 6's revalidation and digest refresh before handoff.
8. STOP. Report the mode, the scope deltas, and the floor verdict; recommend $rite-define.

Mid-flight discipline. Do not replace the interactive review with strategy.md, expand implementation surface without need, grow scope outside the Drift Guard, or score before citing evidence. See reference/anti-patterns.md.

Output

Progress first: run devrites-engine progress, then use the shared completion reply contract (devrites-lib/reference/reply-contract.md). Default success shape:

Done: spec tempered for <slug>; mode <expand|selective|hold-rigor|reduce-to-MVP|skipped-low-stakes>.
Changed: strategy.md, spec.md, decisions.md, assumptions.md
Evidence: dimension floor <band>; reviewer loop <n>; spec readiness + decision coverage re-checked pass
Open: <none | n deferred non-goals>
Next: $rite-define
Record: .devrites/work/<slug>/strategy.md
↻ Hygiene: /clear before $rite-define

If readiness is blocked, use the shared Stopped / blocked form and route Fix: to $rite-clarify for an uncovered decision surface or $rite-spec for an objective spec defect; do not recommend $rite-define. DO NOT plan, slice, or write code here. That's $rite-define and $rite-build.