devrites 3.0.7 → 3.2.0

package/pack/generated/codex/skills/rite-customize/SKILL.md

@@ -12,9 +12,13 @@ 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.
-- When this skill asks for a DevRites specialist or writer agent, **explicitly** spawn the matching Codex custom agent from `.codex/agents/devrites-*.toml` through Codex subagents (`spawn_agent`), then wait for its result and reconcile it as the skill instructs. Do not do the review inline just because the instruction to spawn is embedded here: Codex under-fires embedded spawn/skill instructions (openai/codex #23496), so treat the spawn as required, not optional.
-- The independence of a fresh-context subagent is the point. If Codex genuinely cannot spawn subagents in the current surface, run the documented inline fallback and **label the result an inline fallback, not an independent review**: an inline pass shares the calling context and is weaker evidence.
-- Codex project hooks are installed in `.codex/hooks.json`. Review and trust them with `/hooks` before relying on hook enforcement.
+- **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.
+- For every DevRites specialist or writer dispatch, first call `spawn_agent` with the named `devrites-<role>` custom role. The matching project contract is `.codex/agents/devrites-<role>.toml`.
+- If `spawn_agent` is callable but a named read-only role is unavailable, use generic `explorer` only when the host proves that run has a runtime-enforced read-only sandbox. Tell it to read `.codex/agents/devrites-<role>.toml`, follow its `developer_instructions`, and execute the unchanged packet. A missing read-only custom role is not evidence that spawning is unavailable.
+- Never dispatch generic `worker` for `devrites-slice-wright` unless the host proves that worker run carries exact DevRites identity and the same `.wright-allowlist` enforcement as the named role. Codex reports a generic run as `agent_type=worker`, so the generated global hooks cannot prove that binding. Reject that unsafe rung and use the documented labelled inline wright path with `.reconcile-inline` plus the full reconcile gate.
+- If the host cannot prove the generic explorer is runtime read-only, reject that rung too. Only when no spawn primitive exists or a higher-priority policy rejects a safe spawn may the root run the documented discipline inline. Label it `independence: fallback`, never call it independent, and apply every fallback risk gate. An unbound generic wright or unconfined generic explorer is such a safety rejection, not evidence that no agents exist.
+- 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.
 
 

package/pack/generated/codex/skills/rite-define/SKILL.md

@@ -11,19 +11,23 @@ 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.
-- When this skill asks for a DevRites specialist or writer agent, **explicitly** spawn the matching Codex custom agent from `.codex/agents/devrites-*.toml` through Codex subagents (`spawn_agent`), then wait for its result and reconcile it as the skill instructs. Do not do the review inline just because the instruction to spawn is embedded here: Codex under-fires embedded spawn/skill instructions (openai/codex #23496), so treat the spawn as required, not optional.
-- The independence of a fresh-context subagent is the point. If Codex genuinely cannot spawn subagents in the current surface, run the documented inline fallback and **label the result an inline fallback, not an independent review**: an inline pass shares the calling context and is weaker evidence.
-- Codex project hooks are installed in `.codex/hooks.json`. Review and trust them with `/hooks` before relying on hook enforcement.
+- **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.
+- For every DevRites specialist or writer dispatch, first call `spawn_agent` with the named `devrites-<role>` custom role. The matching project contract is `.codex/agents/devrites-<role>.toml`.
+- If `spawn_agent` is callable but a named read-only role is unavailable, use generic `explorer` only when the host proves that run has a runtime-enforced read-only sandbox. Tell it to read `.codex/agents/devrites-<role>.toml`, follow its `developer_instructions`, and execute the unchanged packet. A missing read-only custom role is not evidence that spawning is unavailable.
+- Never dispatch generic `worker` for `devrites-slice-wright` unless the host proves that worker run carries exact DevRites identity and the same `.wright-allowlist` enforcement as the named role. Codex reports a generic run as `agent_type=worker`, so the generated global hooks cannot prove that binding. Reject that unsafe rung and use the documented labelled inline wright path with `.reconcile-inline` plus the full reconcile gate.
+- If the host cannot prove the generic explorer is runtime read-only, reject that rung too. Only when no spawn primitive exists or a higher-priority policy rejects a safe spawn may the root run the documented discipline inline. Label it `independence: fallback`, never call it independent, and apply every fallback risk gate. An unbound generic wright or unconfined generic explorer is such a safety rejection, not evidence that no agents exist.
+- 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-define: plan from the spec
 
-Read the active feature's `spec.md` and turn it into a buildable workspace: feature
-architecture, approach, a dependency-ordered set of **vertical slices**, traceability, and
-the state cursor. The spec is the WHAT/WHY (from `$rite-spec`); this is the HOW. Splitting
-spec, architecture, plan, tasks, and traceability keeps each file small and phase-owned.
-**No code here.**
+Turn the active feature's `spec.md` into a buildable workspace with architecture, an
+implementation approach, dependency-ordered **vertical slices**, traceability, and a
+state cursor. The spec defines what and why; this phase defines how. Keep spec,
+architecture, plan, tasks, and traceability in their phase-owned files. **Do not write
+code here.**
 
 ## Rules consulted (read on demand from `.agents/skills/devrites-lib/reference/standards/`)
 Pull these via `Read` when shaping the plan:
@@ -36,14 +40,16 @@ Pull these via `Read` when shaping the plan:
 - **Requires a readied spec.** Read the active workspace first; if `.devrites/ACTIVE` is empty,
   the workspace has no `spec.md`, its readiness gate hasn't passed, or any spec-quality
   `checklists/<domain>.md` has an open CRITICAL → **STOP** and tell the user to run
-  `$rite-spec <feature>` first. **DO NOT plan from a missing or unreadied spec.**
-- Prefer existing conventions; ask before adding a dependency or a second design system.
-- **Author section by section, not in one dump.** Write `architecture.md` / `plan.md` one section
-  at a time and pause after each; a section resting on an open design choice or a shaky estimate can
-  be deepened right there with a technique from
+  `$rite-spec <feature>` first. A missing or non-`CLEAR` `decision-coverage.md` routes to
+  `$rite-clarify`. **DO NOT plan from a missing, unreadied, or unclarified spec.**
+- Apply `afk-hitl.md` decision ownership. Prefer conventions; source-check new dependencies or
+  design systems, asking only about licensing/cost/security or explicit policy.
+- **Author one section at a time.** Write `architecture.md` and `plan.md` section by
+  section, pausing after each. For an open design choice or uncertain estimate, use a
+  relevant technique from
   [`elicitation.md`](../devrites-lib/reference/standards/elicitation.md) (Tournament for two viable
   designs, Delphi for the estimate) before it hardens into slices.
-- **Slice count is derived, never dictated.** The number of slices falls out of the work.
+- **Derive the slice count from the work.**
   One per independently-shippable increment, sized by `slicing.md`, every acceptance
   criterion mapped to ≥1 slice. A user-named count is a hint at most: slice logically and,
   if your honest count differs, present it and why. Never pad or compress to hit a figure.
@@ -52,6 +58,9 @@ Pull these via `Read` when shaping the plan:
   crosses many files, don't fake vertical slices. Add a compatibility/adapter slice, migrate
   small green batches, then remove the old path. If a batch cannot stay green, use an
   integration branch plus a final verify slice.
+- **Root writes; drafter proposes.** Use the file-backed fresh-context contract in
+  [`agents.md`](../devrites-lib/reference/standards/agents.md). The root owns architecture
+  decisions, human questions, approval, and every canonical artifact write.
 
 ## Workflow
 0. **Read `.agents/skills/devrites-lib/reference/standards/core.md`:** the always-on operating rules and anti-rationalizations.
@@ -65,30 +74,46 @@ Pull these via `Read` when shaping the plan:
    ```
    If there is no active workspace, no `spec.md`, `spec-skeleton` blocks, or its readiness gate hasn't passed →
    **STOP** and tell the user to run `$rite-spec <feature>` first.
+   If `decision-coverage.md` is absent or does not say `Decision coverage: CLEAR`,
+   **STOP** → `$rite-clarify`.
 1. **Read the spec:** `spec.md` (objective, requirements, acceptance, **placement**,
    design references, gaps/decisions), plus `references.md`, `decisions.md`,
-   `assumptions.md`, **`strategy.md` if present** (the scope mode, deferred / out-of-scope
+   `assumptions.md`, `decision-coverage.md`, **`strategy.md` if present** (the scope mode, deferred / out-of-scope
    register, and pre-mortem risks from `$rite-temper`: cut slices to mitigate the top risks
    and respect the IN/OUT line; map coverage against the **hardened** spec), and
    **`design-brief.md` if the feature touches UI** (the UX/UI contract `$rite-spec` shaped:
    its key states, interaction model, and proof targets drive how UI slices are cut). If a blocking
-   `[NEEDS CLARIFICATION]` remains, stop → `$rite-spec`.
-2. **Decide the architecture + approach** (the HOW the spec deliberately omitted): write
+   `[NEEDS CLARIFICATION]` remains, stop → `$rite-clarify`.
+1a. **Draft from fresh context.** Freeze the planning inputs and dispatch
+   `devrites-plan-drafter` in `define` mode for one atomic candidate bundle:
+   `architecture.md`, `plan.md`, `tasks.md`, and `traceability.md`, including proof mapping.
+   Await and validate `agent-result/v1`. The drafter does not write or ask; any human-owned
+   choice returns to this root context.
+2. **Reconcile and decide the architecture + approach** (the HOW the spec deliberately
+   omitted). Validate the candidate against live seams and the following rules; the root writes
+   accepted content at step 6. Shape
    `architecture.md` for owning layer, boundaries, integration points, data/API/events,
    dependencies, risks, and affected areas; write only the build strategy in `plan.md`.
    Use a
    code-intelligence index if available (see
    `.agents/skills/devrites-lib/reference/standards/tooling.md`) for structure/impact; for the current API or behaviour of
    an external library/framework the architecture will rely on, consult context7 if available.
-   Record significant options in `decisions.md` as `DEC-###` ADR entries.
-   **Deep-modules check**, while sketching the major modules, look for opportunities
+   Record significant options as `DEC-###`. For high-cost/hard-to-reverse boundaries, data
+   models, public contracts, or dependencies, compare ≥2 viable approaches by drivers,
+   trade-offs, and consequences. Specify cross-boundary interfaces for independent work:
+   invariants, I/O, ordering/idempotency, errors, versioning, config, and relevant budgets.
+   **Deep-module check:** while sketching the major modules, look for opportunities
    to extract a **deep module**: a small, stable interface that hides a meaningful chunk
-   of behavior, and is therefore independently testable. A *shallow* module (interface
-   nearly as complex as its implementation) earns nothing; either deepen it or delete
-   it. Where a slice will produce a deep module, confirm with the user which deep
-   modules they want unit-tested in isolation (this informs the slice's "Tests to
-   write/run" field).
-3. **Slice into vertical tasks:** each delivers one observable capability end-to-end and
+   of behavior and is independently testable. A *shallow* module whose interface is
+   nearly as complex as its implementation adds no value; deepen it or delete
+   it. Put independently testable deep-module behavior in the slice's `Tests/proof`;
+   `$rite-vet` confirms the level.
+2a. **Foreseeable-decision sweep.** Inspect questions, assumptions, architecture, dependencies,
+   proof prerequisites, and proposed checkpoints. Search facts and decide reversible technical
+   calls. New product/acceptance/policy/irreversible-risk gaps return to `$rite-clarify`; a build
+   checkpoint survives only for unavailable pre-code evidence or mandatory action-time approval.
+   **Completion:** no known implementation choice is postponed for build to ask later.
+3. **Create vertical tasks:** each delivers one observable capability end to end and
    is verifiable on its own; the **count emerges from the work, not a target number**;
    first slice = thinnest useful end-to-end path; order by dependency (risk-first within a
    tier). For a broad mechanical refactor, use expand → migrate batches → contract instead
@@ -99,12 +124,15 @@ Pull these via `Read` when shaping the plan:
    first, see `devrites-frontend-craft/reference/fullstack.md`). **For UI slices, name which
    of `design-brief.md`'s key states + interaction the slice delivers, and give it a binary
    **Visual acceptance** target (state × viewport × input + target R-id/brief rule)**, so
-   the design contract maps onto slices, not just acceptance criteria.
-4. **Map coverage:** every `AC-###` spec acceptance criterion maps to ≥1 `SLICE-###`
+   the design contract maps to slices as well as acceptance criteria.
+   `Tests/proof` names exact command, cwd, expected signal, prerequisites, and mutable
+   provenance inputs; `$rite-vet` preflights them. Write the portable repository command,
+   never RTK/local wrappers, user-specific absolute paths, or temporary proof trees.
+4. **Map coverage and wiring:** every `AC-###` spec acceptance criterion maps to ≥1 `SLICE-###`
    (`rite-spec/reference/acceptance-criteria.md`); no orphaned criteria, no slice without a
    criterion. Lift covered/backstop `Edge Coverage` rows and resolved `Prohibitions (must-NOT)`
-   rows into `traceability.md` and `test-plan.md`; unresolved rows go to `assumptions.md` with
-   their gate/owner.
+   rows into `traceability.md` and `test-plan.md`; unresolved rows get a gate/owner. Each
+   cross-slice boundary names producer, consumer, invariant, integration step, and proof.
 4a. **Parallel-lane sanity check**: after drafting `tasks.md` but before asking for plan
    approval, run the advisory lane planner:
    ```bash
@@ -121,7 +149,7 @@ Pull these via `Read` when shaping the plan:
    S="$(cat .devrites/ACTIVE 2>/dev/null)"
    devrites-engine coverage "$S" > ".devrites/work/$S/traceability.md"
    ```
-5. **Complexity & deviations gate:** justify anything off DevRites defaults (new dep,
+5. **Complexity and deviations gate:** justify anything outside DevRites defaults (new dep,
    extra abstraction, second design system) in the plan; if you can't justify it, simplify.
    **Principles conformance:** read `.devrites/principles.md` (if present) and confirm the
    approach honors every declared invariant. A plan that conflicts with one is not "a deviation
@@ -137,11 +165,12 @@ Pull these via `Read` when shaping the plan:
    S="$(cat .devrites/ACTIVE 2>/dev/null)"
    devrites-engine analyze "$S"
    ```
-7. **Readiness gate** (bottom of plan-template): every acceptance criterion covered by a
-   slice, dependency order acyclic + risk-first, no unjustified deviation, rollback for
-   every destructive/migration step. **Stop and confirm** before code. Render the review-before-code
+7. **Readiness gate** (plan-template): require CLEAR coverage, complete acceptance and
+   cross-slice wiring/proof, risk-first acyclic order, justified deviations, rollback, and a
+   closed decision sweep. **Stop and confirm** before code. Render the review-before-code
    digest first: `Intent` (one sentence from the spec), `Done means` (acceptance coverage x/y),
-   `Plan sanity` (slice count + riskiest boundary/gate), and `Build exactly this?` (yes → approve;
+   `Plan sanity` (slice count + riskiest boundary/gate), `Expected build interruptions`
+   (`none` or only justified action-time gates), and `Build exactly this?` (yes → approve;
    no → `$rite-plan revise`). When the human confirms the plan, write `Plan approved: <iso>` to
    `state.md` (see [state-workspace](../rite-spec/reference/state-workspace.md)); `$rite-build`
    checks this exists before building.
@@ -153,7 +182,8 @@ Use the canonical slice grammar in
 Every slice must satisfy that complete field set; phase-specific gate details live in
 [`reference/gates.md`](reference/gates.md).
 
-> **Mid-flight discipline.** When tempted to skip vertical slicing, coverage mapping, or dependency-order discipline: see [`anti-patterns`](reference/anti-patterns.md) (Common Rationalizations + Red Flags). Load it the moment you reach for the excuse.
+> **Mid-flight discipline.** Do not skip vertical slicing, coverage mapping, or
+> dependency ordering. See [`anti-patterns`](reference/anti-patterns.md).
 
 ## Output
 

package/pack/generated/codex/skills/rite-define/reference/gates.md

@@ -1,12 +1,10 @@
 # Gate taxonomy: advisory · validating · blocking · escalating
 
-DevRites uses a four-gate model for HITL pauses, adapted from the regulated-agentic-workflow
-governance pattern. Picking the right gate for each `Mode: HITL` slice is the difference
-between a workflow that catches real risk and one that becomes a review queue.
+DevRites uses four HITL gates, adapted from the regulated-agentic-workflow governance
+pattern. Choose the gate for each `Mode: HITL` slice by its risk and review needs.
 
-> **Default failure mode:** marking every HITL slice as `blocking`. The same reviewer ends
-> up approving low-stakes and high-stakes items at the same priority and the gate becomes
-> a queue. Mix gate types per slice; most plans use 2-3 of the 4.
+> **Do not mark every HITL slice as `blocking`.** That gives low-stakes and high-stakes
+> items the same priority. Most plans use 2 or 3 gate types.
 
 ## The four gates
 
@@ -39,7 +37,7 @@ validating question; the feature does not seal until the entry is resolved. An o
 **Example:** "Schema migration adds a non-null column with a default. Backfill plan is
 recorded; reviewer should confirm the default is the right one for archived rows."
 
-**SLA:** `4h`: the work continues but the validating queue should drain within hours,
+**SLA:** `4h`: the work continues, but the validating queue should clear within hours,
 not days.
 
 ### blocking
@@ -56,8 +54,8 @@ STOPs. The slice is not built until `$rite-resolve` lands.
 - Auth/authz boundary change.
 - Public API break.
 - Spec drift that changes acceptance criteria.
-- Tests / types / lint are red and the agent cannot tell whether the slice's contract is
-  wrong or the failing code is.
+- Bounded debug recovery proved the remaining red test/type/lint failure is a genuine
+  product-contract ambiguity the human must decide.
 
 **SLA:** `15m`: synchronous gates demand fast turnaround; otherwise treat the work as
 genuinely blocked and re-plan around it.
@@ -79,7 +77,9 @@ it's `blocking`.
 
 ## Picking the gate
 
-Apply this decision tree per HITL slice:
+First apply `afk-hitl.md` decision ownership: an objective implementation/tooling failure
+or reversible technical choice is agent work and gets no human gate. Then apply this
+decision tree per HITL slice:
 
 1. **Can the slice ship safely without the answer?**
    - Yes → `advisory`.
@@ -117,8 +117,9 @@ defaults and the always-pause rules:
 | `[advisory, validating]` | log + proceed | build + queue | pause | pause |
 | `[advisory, validating, blocking]` | log + proceed | build + queue | log + proceed* | pause |
 
-\* but **never** for destructive migrations, auth/authz boundary changes, public API
-breaks, or red tests/types/lint. Those always pause. See
+\* but **never** for destructive migrations, auth/authz boundary changes, or public API
+breaks. Red tests/types/lint remain hard build gates and must clear bounded recovery before
+the next slice; only a resulting human-owned ambiguity becomes a pause. See
 [`.agents/skills/devrites-lib/reference/standards/afk-hitl.md`](../../devrites-lib/reference/standards/afk-hitl.md) for the irreversible-risk
 list.
 
@@ -127,8 +128,7 @@ shortcut.
 
 ## Anti-patterns
 
-- **One gate for everything.** Validates becomes a queue, blocks all work behind one
-  reviewer. Pick gates per slice.
+- **One gate for everything.** This puts all work behind one reviewer. Pick gates per slice.
 - **Marking a destructive migration `validating` to "keep the loop moving".** Destructive
   work is `blocking` regardless of the urge to ship.
 - **`advisory` as a synonym for "I'm not sure but I don't want to ask".** If the slice
@@ -143,7 +143,8 @@ shortcut.
 Mode: HITL
 Gate: blocking
 SLA: 15m
-Checkpoint: Confirm (user_id, created_at) composite index choice vs two single-col indexes.
+Checkpoint: Approve irreversible deletion of legacy records after the dry-run count exists;
+that evidence cannot exist before the migration rehearsal.
 Blocked by: SLICE-002
 ...
 ```

package/pack/generated/codex/skills/rite-define/reference/plan-template.md

@@ -1,12 +1,11 @@
 # `plan.md` template
 
-Write **how** to build what `spec.md` defines. The plan is **living, not sacred**:
-`$rite-plan` repairs it when reality disagrees. This is where technology choices live
-(they're banned from the spec).
+Write **how** to build what `spec.md` defines. Use `$rite-plan` when evidence requires
+an update. Technology choices belong here, not in the spec.
 
 ```markdown
 # Plan: <Feature>
-Spec: ./spec.md   Date: <date>
+Spec: ./spec.md   Decision coverage: ./decision-coverage.md   Date: <date>
 
 ## Summary
 1–2 sentences: the primary requirement + the chosen approach.
@@ -17,7 +16,8 @@ Spec: ./spec.md   Date: <date>
 - Storage / data: <...>
 - Testing tools + commands: <from spec "Commands discovered">
 - Target / platform / constraints: <...>
-- `[NEEDS CLARIFICATION: ...]` for any unknown that affects the approach.
+- Any product/constraint unknown that affects the approach routes to `$rite-clarify`;
+  do not carry `[NEEDS CLARIFICATION]` into an approvable plan.
 
 ## Global constraints
 Project-wide requirements from the spec that every slice implicitly includes — version
@@ -26,6 +26,8 @@ values **verbatim from spec.md** (a paraphrased constraint drifts by the time sl
 
 ## Approach
 The strategy in a few sentences. Why this over the alternatives considered.
+For high-cost/hard-to-reverse boundaries, models, contracts, or dependencies compare ≥2
+viable `Option · Drivers · Trade-offs · Consequence`; skip theater for routine choices.
 
 ## Slice strategy
 How the feature is split into vertical `SLICE-###` increments, why the order is
@@ -36,9 +38,9 @@ green, or name the integration branch + final verify slice.
 ## Architecture decisions
 Key decisions + rationale (mirror into decisions.md). New pattern vs reuse — prefer
 reuse of existing project conventions. Record architecture as invariants, not scaffolding:
-include a decision only when two implementing slices could otherwise choose incompatibly. For each
-medium+ decision, include `Binds:` (what future work must follow) and `Prevents:` (the divergence
-or failure it avoids).
+include a decision only when slices could choose incompatibly. Medium+ decisions include
+`Binds:` and `Prevents:`. Cross-boundary interfaces name invariants, I/O, ordering/idempotency,
+errors, versioning, config, and relevant budgets.
 
 ## Dependency graph
 What must exist before what (text is fine):
@@ -55,6 +57,9 @@ slices above the cut, and no slice above it depends on one below.
 ## Validation strategy
 After which slices to run tests / build / browser proof. For UI slices, name the visual
 acceptance targets from `design-brief.md`, not a generic "looks good" check.
+Each proof names exact command/cwd/signal, prerequisites, and mutable provenance inputs.
+Use portable repository commands; runtime wrappers and local absolute/temp paths do not
+belong in a durable plan.
 
 **Key links** — the wiring the assembled feature must exhibit, one row each:
 `<from> → <to> via <mechanism>` (route → handler via registration; producer → consumer via
@@ -81,14 +86,19 @@ Framework/library docs to consult (triggers devrites-source-driven). Record URLs
 decisions.md / evidence.md when used.
 
 ## Readiness gate  *(must pass before $rite-build)*
+- [ ] `decision-coverage.md` says `Decision coverage: CLEAR`
 - [ ] Every spec acceptance criterion is covered by a slice
 - [ ] Dependency order is acyclic and risk-first
 - [ ] An `MVP cut` is named, self-contained (ACs above the cut proven above the cut, no dependency reaching below), and marks a genuinely shippable scope
 - [ ] No unjustified deviation remains in the complexity gate
 - [ ] Rollback exists for every destructive / migration step
 - [ ] Every `Mode: HITL` slice has `Gate`, `SLA`, and `Checkpoint` populated
+- [ ] Human-owned choices are resolved; surviving checkpoints need unavailable pre-code
+      evidence or action-time approval
+- [ ] Slice proof commands, cwd, prerequisites, and provenance inputs are portable and preflightable
 - [ ] Every UI slice names `Design brief states` and binary `Visual acceptance`
 - [ ] `Key links` rows cover every cross-slice wiring (or state `none`)
+- [ ] Cross-boundary contracts name producer, consumer, invariants/errors/order, and proof
 - [ ] Any wide mechanical refactor is sliced expand → migrate batches → contract, with green migrate batches or an integration branch + final verify slice
 - [ ] No `Gate: blocking` slice is implicitly chained behind an AFK slice without surfacing the dependency
 ```

package/pack/generated/codex/skills/rite-doctor/SKILL.md

@@ -11,9 +11,13 @@ 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.
-- When this skill asks for a DevRites specialist or writer agent, **explicitly** spawn the matching Codex custom agent from `.codex/agents/devrites-*.toml` through Codex subagents (`spawn_agent`), then wait for its result and reconcile it as the skill instructs. Do not do the review inline just because the instruction to spawn is embedded here: Codex under-fires embedded spawn/skill instructions (openai/codex #23496), so treat the spawn as required, not optional.
-- The independence of a fresh-context subagent is the point. If Codex genuinely cannot spawn subagents in the current surface, run the documented inline fallback and **label the result an inline fallback, not an independent review**: an inline pass shares the calling context and is weaker evidence.
-- Codex project hooks are installed in `.codex/hooks.json`. Review and trust them with `/hooks` before relying on hook enforcement.
+- **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.
+- For every DevRites specialist or writer dispatch, first call `spawn_agent` with the named `devrites-<role>` custom role. The matching project contract is `.codex/agents/devrites-<role>.toml`.
+- If `spawn_agent` is callable but a named read-only role is unavailable, use generic `explorer` only when the host proves that run has a runtime-enforced read-only sandbox. Tell it to read `.codex/agents/devrites-<role>.toml`, follow its `developer_instructions`, and execute the unchanged packet. A missing read-only custom role is not evidence that spawning is unavailable.
+- Never dispatch generic `worker` for `devrites-slice-wright` unless the host proves that worker run carries exact DevRites identity and the same `.wright-allowlist` enforcement as the named role. Codex reports a generic run as `agent_type=worker`, so the generated global hooks cannot prove that binding. Reject that unsafe rung and use the documented labelled inline wright path with `.reconcile-inline` plus the full reconcile gate.
+- If the host cannot prove the generic explorer is runtime read-only, reject that rung too. Only when no spawn primitive exists or a higher-priority policy rejects a safe spawn may the root run the documented discipline inline. Label it `independence: fallback`, never call it independent, and apply every fallback risk gate. An unbound generic wright or unconfined generic explorer is such a safety rejection, not evidence that no agents exist.
+- 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.
 
 

package/pack/generated/codex/skills/rite-dogfood/SKILL.md

@@ -12,9 +12,13 @@ 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.
-- When this skill asks for a DevRites specialist or writer agent, **explicitly** spawn the matching Codex custom agent from `.codex/agents/devrites-*.toml` through Codex subagents (`spawn_agent`), then wait for its result and reconcile it as the skill instructs. Do not do the review inline just because the instruction to spawn is embedded here: Codex under-fires embedded spawn/skill instructions (openai/codex #23496), so treat the spawn as required, not optional.
-- The independence of a fresh-context subagent is the point. If Codex genuinely cannot spawn subagents in the current surface, run the documented inline fallback and **label the result an inline fallback, not an independent review**: an inline pass shares the calling context and is weaker evidence.
-- Codex project hooks are installed in `.codex/hooks.json`. Review and trust them with `/hooks` before relying on hook enforcement.
+- **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.
+- For every DevRites specialist or writer dispatch, first call `spawn_agent` with the named `devrites-<role>` custom role. The matching project contract is `.codex/agents/devrites-<role>.toml`.
+- If `spawn_agent` is callable but a named read-only role is unavailable, use generic `explorer` only when the host proves that run has a runtime-enforced read-only sandbox. Tell it to read `.codex/agents/devrites-<role>.toml`, follow its `developer_instructions`, and execute the unchanged packet. A missing read-only custom role is not evidence that spawning is unavailable.
+- Never dispatch generic `worker` for `devrites-slice-wright` unless the host proves that worker run carries exact DevRites identity and the same `.wright-allowlist` enforcement as the named role. Codex reports a generic run as `agent_type=worker`, so the generated global hooks cannot prove that binding. Reject that unsafe rung and use the documented labelled inline wright path with `.reconcile-inline` plus the full reconcile gate.
+- If the host cannot prove the generic explorer is runtime read-only, reject that rung too. Only when no spawn primitive exists or a higher-priority policy rejects a safe spawn may the root run the documented discipline inline. Label it `independence: fallback`, never call it independent, and apply every fallback risk gate. An unbound generic wright or unconfined generic explorer is such a safety rejection, not evidence that no agents exist.
+- 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.
 
 

package/pack/generated/codex/skills/rite-explain/SKILL.md

@@ -12,9 +12,13 @@ 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.
-- When this skill asks for a DevRites specialist or writer agent, **explicitly** spawn the matching Codex custom agent from `.codex/agents/devrites-*.toml` through Codex subagents (`spawn_agent`), then wait for its result and reconcile it as the skill instructs. Do not do the review inline just because the instruction to spawn is embedded here: Codex under-fires embedded spawn/skill instructions (openai/codex #23496), so treat the spawn as required, not optional.
-- The independence of a fresh-context subagent is the point. If Codex genuinely cannot spawn subagents in the current surface, run the documented inline fallback and **label the result an inline fallback, not an independent review**: an inline pass shares the calling context and is weaker evidence.
-- Codex project hooks are installed in `.codex/hooks.json`. Review and trust them with `/hooks` before relying on hook enforcement.
+- **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.
+- For every DevRites specialist or writer dispatch, first call `spawn_agent` with the named `devrites-<role>` custom role. The matching project contract is `.codex/agents/devrites-<role>.toml`.
+- If `spawn_agent` is callable but a named read-only role is unavailable, use generic `explorer` only when the host proves that run has a runtime-enforced read-only sandbox. Tell it to read `.codex/agents/devrites-<role>.toml`, follow its `developer_instructions`, and execute the unchanged packet. A missing read-only custom role is not evidence that spawning is unavailable.
+- Never dispatch generic `worker` for `devrites-slice-wright` unless the host proves that worker run carries exact DevRites identity and the same `.wright-allowlist` enforcement as the named role. Codex reports a generic run as `agent_type=worker`, so the generated global hooks cannot prove that binding. Reject that unsafe rung and use the documented labelled inline wright path with `.reconcile-inline` plus the full reconcile gate.
+- If the host cannot prove the generic explorer is runtime read-only, reject that rung too. Only when no spawn primitive exists or a higher-priority policy rejects a safe spawn may the root run the documented discipline inline. Label it `independence: fallback`, never call it independent, and apply every fallback risk gate. An unbound generic wright or unconfined generic explorer is such a safety rejection, not evidence that no agents exist.
+- 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.
 
 

package/pack/generated/codex/skills/rite-frame/SKILL.md

@@ -11,9 +11,13 @@ 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.
-- When this skill asks for a DevRites specialist or writer agent, **explicitly** spawn the matching Codex custom agent from `.codex/agents/devrites-*.toml` through Codex subagents (`spawn_agent`), then wait for its result and reconcile it as the skill instructs. Do not do the review inline just because the instruction to spawn is embedded here: Codex under-fires embedded spawn/skill instructions (openai/codex #23496), so treat the spawn as required, not optional.
-- The independence of a fresh-context subagent is the point. If Codex genuinely cannot spawn subagents in the current surface, run the documented inline fallback and **label the result an inline fallback, not an independent review**: an inline pass shares the calling context and is weaker evidence.
-- Codex project hooks are installed in `.codex/hooks.json`. Review and trust them with `/hooks` before relying on hook enforcement.
+- **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.
+- For every DevRites specialist or writer dispatch, first call `spawn_agent` with the named `devrites-<role>` custom role. The matching project contract is `.codex/agents/devrites-<role>.toml`.
+- If `spawn_agent` is callable but a named read-only role is unavailable, use generic `explorer` only when the host proves that run has a runtime-enforced read-only sandbox. Tell it to read `.codex/agents/devrites-<role>.toml`, follow its `developer_instructions`, and execute the unchanged packet. A missing read-only custom role is not evidence that spawning is unavailable.
+- Never dispatch generic `worker` for `devrites-slice-wright` unless the host proves that worker run carries exact DevRites identity and the same `.wright-allowlist` enforcement as the named role. Codex reports a generic run as `agent_type=worker`, so the generated global hooks cannot prove that binding. Reject that unsafe rung and use the documented labelled inline wright path with `.reconcile-inline` plus the full reconcile gate.
+- If the host cannot prove the generic explorer is runtime read-only, reject that rung too. Only when no spawn primitive exists or a higher-priority policy rejects a safe spawn may the root run the documented discipline inline. Label it `independence: fallback`, never call it independent, and apply every fallback risk gate. An unbound generic wright or unconfined generic explorer is such a safety rejection, not evidence that no agents exist.
+- 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.
 
 

package/pack/generated/codex/skills/rite-frame/reference/failure-modes.md

@@ -1,9 +1,8 @@
-# The four failure modes: mapped to their DevRites cure
+# Four failure modes and their DevRites controls
 
-The lens behind `rite-frame`. Each mode is a thing LLMs reliably get wrong; each already
-has a cure somewhere in the pack. The value of naming them here is a **portable
-vocabulary**: a fast self-check you can run on ad-hoc or express-lane work that never
-reaches the gates where those cures normally fire.
+`rite-frame` checks four common LLM failure modes. Existing DevRites controls address
+each one. Use these names for quick checks on ad hoc or express-lane work that does not
+reach the normal gates.
 
 ## The map
 
@@ -11,18 +10,17 @@ reaches the gates where those cures normally fire.
 |---|---|---|---|
 | 1 | **Silent assumption** | A value, contract, edge case, or interpretation was *guessed* and run with: no note, no question. The ask had two readings and one was picked silently. | `core.md` #2 (no silent assumptions) / #3 (no guessing through confusion); the Spec Drift Guard; `devrites-doubt` for a single high-stakes call; `devrites-interview` when the ask is underspecified. |
 | 2 | **Overcomplication** | An abstraction, flag, config knob, or indirection nobody asked for. 200 lines where 50 would do. A single-use factory. Defensive `try/catch` + null checks inside trusted code. A dependency added where an in-repo option exists. | `coding-style.md` (simplicity, reuse-first), `patterns.md` (avoid over-engineering), `devrites-audit simplify` (deletion test, Chesterton's Fence), the "over-defensive guarding is slop" anti-pattern. |
-| 3 | **Out-of-scope edit** | Touched code, comments, or formatting outside the ask. A "while I'm here" refactor. Renamed something orthogonal. Reflowed an import block. | `core.md` #7 (feature scope only), `touched-files.md` (the recorded diff boundary), `devrites-engine reconcile` (hard-stops a source file changed outside the claimed set), the "it's only a small refactor" anti-pattern. |
+| 3 | **Out-of-scope edit** | Touched code, comments, or formatting outside the ask. A "while I'm here" refactor. Renamed something orthogonal. Reflowed an import block. | `core.md` #7 (feature scope only), root-owned `.wright-allowlist` (the authorized write boundary), `touched-files.md` (the resulting diff record), `devrites-engine reconcile` (hard-stops a source file outside the captured allowlist), the "it's only a small refactor" anti-pattern. |
 | 4 | **Unverifiable goal** | "It works" with no command, no output, no test. A success criterion that can't be false ("make it better"). A tautological test that passes no matter what. | `rite-spec/reference/acceptance-criteria.md` (measurable acceptance), `core.md` #6 (evidence over confidence), `testing.md` (assertion strength, see it fail first), the TDD wright. |
 
-The first three are the **diseases** Karpathy named; the fourth is the **leverage**: a
-falsifiable criterion is what lets the model loop to done without a human in the loop every
-few minutes. FRAME front-loads mode 4 so the AUDIT of modes 1-3 has something to check
-against.
+Karpathy named the first three. The fourth supplies the falsifiable criterion needed
+for an agent to work without repeated human confirmation. FRAME defines that criterion
+before auditing the other three modes.
 
-## FRAME: the imperative → falsifiable reframe
+## FRAME: rewrite an imperative as a falsifiable condition
 
-The move is to rewrite the *verb* as a *condition that can be false*. If you can't, the ask
-is ambiguous. That's mode 1, caught before the diff instead of after.
+Rewrite the request as a condition that can be false. If that is impossible, the
+request is ambiguous and triggers mode 1 before any diff exists.
 
 | Imperative ask | Weak (still a wish) | Falsifiable criterion + verify |
 |---|---|---|
@@ -33,8 +31,8 @@ is ambiguous. That's mode 1, caught before the diff instead of after.
 | "Handle errors" | "add error handling" | "On `<failure F>` the system fails closed with `<observable>`; a test forces F." → that test |
 | "Improve the UX" | "make it nicer" | *(no falsifiable check → ambiguous → `devrites-ux-shape` / ask which states/flows)* |
 
-A criterion passes the bar when a reviewer could run the verify command and get a clear
-pass or fail without asking you what you meant. If they'd have to ask, sharpen it.
+A criterion is ready when a reviewer can run the verification command and get a clear
+pass or fail without asking what it means.
 
 ## AUDIT: worked finding lines
 
@@ -49,20 +47,20 @@ utils.ts:5  FYI        mode 3 (scope): reflowed an unrelated import block. → r
 sum.spec:12 Important  mode 4 (unverifiable): test asserts `mock.called`, not the result — passes if the fn is empty. → assert the value; see it fail first (testing.md).
 ```
 
-Clean lanes say so explicitly (`1 assumption: clean`) so the audit is a positive
-statement, not a silent skip.
+Record clean modes explicitly (`1 assumption: clean`) so an empty result is not mistaken
+for a skipped check.
 
-## How this differs from the heavier tools (so you don't reach past it)
+## Related tools
 
 - **`devrites-doubt`:** adversarial *fresh-context subagent* pre-mortem on **one** decision.
-  Heavier, independent, anti-anchoring. rite-frame is *self-applied* and covers the four modes
-  broadly; escalate a single load-bearing claim to doubt.
+  It is independent and avoids anchoring. rite-frame is self-applied and covers all four
+  modes; use doubt for one consequential claim.
 - **`devrites-audit <axis>`:** *fresh-context subagent* review of an **active feature's** diff
   on one axis (security / perf / simplify). Needs a `.devrites/` workspace. rite-frame needs no
-  workspace and runs in your own context: cheaper, less independent.
+  workspace and runs in the current context, with less independence.
 - **`$rite-review` · `$rite-seal`:** the **gates**: parallel reviewer fan-out, blocking
   severities, written verdict. rite-frame is the inline reflex for work that never reaches a
-  gate; it does not replace one. If the work is a real feature, the gate is the answer.
+  gate and does not replace one.
 
-The point is not redundancy. It's a tiered ladder. rite-frame is the cheapest rung: a lens
-you run yourself in seconds. Climb to doubt / audit / the gates as the stakes rise.
+Use rite-frame for the quickest self-check. Use doubt, audit, or the full gates when the
+work needs independent or feature-level review.

package/pack/generated/codex/skills/rite-handoff/SKILL.md

@@ -12,9 +12,13 @@ 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.
-- When this skill asks for a DevRites specialist or writer agent, **explicitly** spawn the matching Codex custom agent from `.codex/agents/devrites-*.toml` through Codex subagents (`spawn_agent`), then wait for its result and reconcile it as the skill instructs. Do not do the review inline just because the instruction to spawn is embedded here: Codex under-fires embedded spawn/skill instructions (openai/codex #23496), so treat the spawn as required, not optional.
-- The independence of a fresh-context subagent is the point. If Codex genuinely cannot spawn subagents in the current surface, run the documented inline fallback and **label the result an inline fallback, not an independent review**: an inline pass shares the calling context and is weaker evidence.
-- Codex project hooks are installed in `.codex/hooks.json`. Review and trust them with `/hooks` before relying on hook enforcement.
+- **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.
+- For every DevRites specialist or writer dispatch, first call `spawn_agent` with the named `devrites-<role>` custom role. The matching project contract is `.codex/agents/devrites-<role>.toml`.
+- If `spawn_agent` is callable but a named read-only role is unavailable, use generic `explorer` only when the host proves that run has a runtime-enforced read-only sandbox. Tell it to read `.codex/agents/devrites-<role>.toml`, follow its `developer_instructions`, and execute the unchanged packet. A missing read-only custom role is not evidence that spawning is unavailable.
+- Never dispatch generic `worker` for `devrites-slice-wright` unless the host proves that worker run carries exact DevRites identity and the same `.wright-allowlist` enforcement as the named role. Codex reports a generic run as `agent_type=worker`, so the generated global hooks cannot prove that binding. Reject that unsafe rung and use the documented labelled inline wright path with `.reconcile-inline` plus the full reconcile gate.
+- If the host cannot prove the generic explorer is runtime read-only, reject that rung too. Only when no spawn primitive exists or a higher-priority policy rejects a safe spawn may the root run the documented discipline inline. Label it `independence: fallback`, never call it independent, and apply every fallback risk gate. An unbound generic wright or unconfined generic explorer is such a safety rejection, not evidence that no agents exist.
+- 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.