@orkestrel/scaffold 0.0.77 → 0.0.79

package/dist/host/agents/skills/enterprise-bootstrap/SKILL.md

@@ -11,7 +11,7 @@ description: >-
   scales, or must look professional rather than like stock Bootstrap. Covers
   aesthetics, typography, color modes, design tokens, elevation, finishing
   details, accessibility (WCAG 2.2 AA), responsive layout, and enterprise app
-  patterns. The `orkestrel-polish-surface` skill owns a requested verdict,
+  patterns. The `orkestrel-polish` skill owns a requested verdict,
   round, or campaign over a surface that already renders, including a review
   that changes nothing; in that campaign's fix units, use this skill for
   Bootstrap craft.
@@ -58,7 +58,7 @@ as a runtime producer is the exception, and only for the property that lookup na
 4. **Write framework-neutral markup** — semantic HTML plus Bootstrap classes. Wire behavior with the project's stack; in an SPA prefer framework-native Bootstrap wrappers over competing DOM ownership ([bootstrap-reference.md](references/bootstrap-reference.md) → JavaScript lifecycle).
 5. **Keep this folder intact** so its relative links resolve. Install or vendor it wherever the tooling looks for skills; the paths are tooling-specific, the content is not.
 6. **Use the installed Bootstrap.** Otherwise take the pinned CDN example from [bootstrap-reference.md](references/bootstrap-reference.md) → Quick start. Do not upgrade dependencies as a side effect of a visual pass.
-7. **Apply this skill** to the work in its frontmatter. For a requested verdict, round, or campaign over an already rendering surface, use `orkestrel-polish-surface`; use this skill for the campaign's Bootstrap fixes.
+7. **Apply this skill** to the work in its frontmatter. For a requested verdict, round, or campaign over an already rendering surface, use `orkestrel-polish`; use this skill for the campaign's Bootstrap fixes.
 
 ---
 

package/dist/host/agents/skills/enterprise-bootstrap/references/inspection.md

@@ -257,7 +257,7 @@ a pass label alone.
 
 Use captures for these judgments, not synthetic negative controls. Record the criterion, capture,
 viewport/theme/state, finding, and disposition. A missing capture leaves the judgment open. For a
-requested review round or campaign, use `orkestrel-polish-surface` rather than creating one here.
+requested review round or campaign, use `orkestrel-polish` rather than creating one here.
 
 - **Task and hierarchy:** the main information and action lead; supporting content remains readable;
   labels, semantics, and destructive rank match the work. Check a grayscale view as a hierarchy aid.

package/dist/host/agents/skills/{orkestrel-align-packages → orkestrel-align}/SKILL.md

@@ -1,20 +1,13 @@
 ---
-name: orkestrel-align-packages
+name: orkestrel-align
 description: Audit and improve how two or more Orkestrel packages, or their core, server, browser, and app environments, fit together. Use for coordinated package-stack refactors, cross-package extraction, developer-ergonomics reviews, end-to-end or live integration testing, dependency and guide alignment, and fleet/package-manager campaigns. Preserve host-independent core boundaries, update dependents topologically, and use the package-hardening workflow for each implementation unit.
 ---
 
 # Align Orkestrel packages
 
-## Load authority
+## Read
 
-Read the current authority in this order:
-
-1. `AGENTS.md` and applicable `.claude/rules/*.md`.
-2. `.agents/orchestration.md` for orchestration.
-3. The `integration.md` and `fleet.md` references selected later in this section.
-4. Relevant package guides, `guides/scaffold.md`, and the configured Orkestrel specialist.
-
-Explicit user scope wins.
+`AGENTS.md` § Authority and loading names the files every unit reads. This skill adds the reference selected in this section, `guides/scaffold.md`, and the `orkestrel` role's catalog table.
 
 Use live manifests, lockfiles, installed declarations, guides, branches, and consumer code. Never trust remembered package versions, dependency ranges, or catalog entries.
 
@@ -22,7 +15,7 @@ Load [integration.md](references/integration.md) for cross-package or cross-envi
 
 Load [fleet.md](references/fleet.md) for version/range/guide drift, dependency blast radius, consumer-artifact validation, or campaign state.
 
-Invoke `$orkestrel-harden-package` for each package implementation unit. This skill owns cross-package decisions; the hardening skill owns the complete work inside one package.
+Invoke `$orkestrel-harden` for each package implementation unit. This skill owns cross-package decisions; the hardening skill owns the complete work inside one package.
 
 ## Execute the campaign
 
@@ -32,11 +25,11 @@ Invoke `$orkestrel-harden-package` for each package implementation unit. This sk
 4. **Assign ownership.** Keep core host-independent. Keep server free of browser assumptions and browser free of Node assumptions. Keep product policy in applications.
 5. **Judge downstream friction.** Treat consumer gymnastics as evidence, not automatic proof. Decide whether the consumer misused a coherent API or exposed a general upstream defect.
 6. **Design in dependency order.** Update guides and types first in the lowest owning package, then update every affected consumer atomically. Do not add compatibility shims.
-7. **Harden each unit.** Apply `$orkestrel-harden-package` to every touched package with the exact owned scope and acceptance criteria.
+7. **Harden each unit.** Apply `$orkestrel-harden` to every touched package with the exact owned scope and acceptance criteria.
 8. **Prove the round trip.** Put integration coverage at the highest package that can exercise the real composed behavior. Retain focused deterministic coverage in lower packages.
 9. **Validate local changes safely.** When necessary, use built artifacts in an isolated consumer and prove the resolved graph. Restore temporary manifest/lockfile state.
 10. **Verify topologically.** Run local gates for touched packages from dependencies to dependents and inspect relevant generated outputs.
-11. **Review the campaign.** Run the adversarial pass plus a mechanical checker on each implementation unit, then one cross-package architecture and ergonomics review over the whole graph.
+11. **Review the campaign.** Run the review `.agents/orchestration.md` § Size gate names on the integrated result, with one cross-package architecture and ergonomics pass over the whole graph. Do not review each unit separately.
 
 ## Accept the result
 

package/dist/host/agents/skills/{orkestrel-align-packages → orkestrel-align}/agents/openai.yaml

@@ -1,4 +1,4 @@
 interface:
   display_name: 'Align Orkestrel Packages'
   short_description: 'Align package boundaries, ergonomics, and readiness'
-  default_prompt: 'Use $orkestrel-align-packages to audit and improve this Orkestrel package stack.'
+  default_prompt: 'Use $orkestrel-align to audit and improve this Orkestrel package stack.'

package/dist/host/agents/skills/{orkestrel-align-packages → orkestrel-align}/references/fleet.md

@@ -4,15 +4,13 @@
 
 For every relevant repository, record:
 
-- local package name/version;
-- branch and relation to `origin/main`;
-- dirty paths and owning session;
-- declared and resolved `@orkestrel/*` dependencies, peers, and dev dependencies;
-- vendored guide freshness;
+- `scaffold audit --json` run in that repository: its `releases` measure every declared `@orkestrel/*` range against the registry, and its `findings` name vendored and mirror drift;
+- `git status --porcelain -b`: the branch, its relation to `origin/main`, and the dirty paths, with the owning session;
+- the lockfile's resolved `@orkestrel/*` versions;
 - supported runtime and package environments;
-- local gate/package state.
+- local gate state.
 
-Use npm, git, manifests, lockfiles, installed packages, canonical guides, and current scaffold authority. Do not copy a package catalog or burned-version list into this skill.
+Run `scaffold catalog --json` once for the fleet table (`membership.entries`) and the layer order. Do not copy a package catalog or burned-version list into this skill.
 
 For `0.0.x`, verify the actual range behavior rather than assuming ordinary caret semantics. Record exact intended ranges and resolved versions.
 

package/dist/host/agents/skills/{orkestrel-build-application → orkestrel-build}/SKILL.md

@@ -1,32 +1,22 @@
 ---
-name: orkestrel-build-application
+name: orkestrel-build
 description: Wire, extend, or harden Orkestrel `app/core`, `app/browser`, and `app/server` environments — the environment's contracts, boundaries, entries, and host proofs, never the product designed inside one. Use for app-only or mixed src/app workspaces, app environment isolation, Vue browser entries, Node server entries, app aliases, configs, scripts, and tests, cross-environment contracts, and application guide parity. Do not use it to redesign an application's routes, screens, or domain behavior; take that to a design round and return here to wire what it decides.
 ---
 
 # Build an Orkestrel application
 
-## Load authority
+## Read
 
-Read the current files in this order:
-
-1. `AGENTS.md`.
-2. `.claude/rules/application.md` for composition, entries, manifest safety, and
-   lifecycle; `.claude/rules/workspace.md` for environments, aliases, configuration, and
-   scripts; `.claude/rules/architecture.md` for placement and barrels;
-   `.claude/rules/tests.md` for test law; `.claude/rules/documentation.md` for parity;
-   plus every other rule the files in scope select.
-3. `guides/README.md`, the governing application guide, and `ROADMAP.md` when present.
-4. The authoritative `*/types.ts` of each selected environment, the root
-   `tsconfig.json` and `vite.config.ts`, the manifest, and `configs/app`.
-
-Those rules are the contract; this skill is only the workflow. Where a step names a law,
-read the law rather than this summary of it.
+`AGENTS.md` § Authority and loading names the files every unit reads. This skill adds the
+authoritative `*/types.ts` of each selected environment, the root `tsconfig.json` and
+`vite.config.ts`, the manifest, and `configs/app`. Where a step names a law, read the law rather
+than this summary of it.
 
 ## Select the environments
 
 `--src` selects published src environments and `--app` selects private app environments.
-The selections are independent, at least one is required, each offers `core`, `browser`,
-and `server`, and there is no `--surfaces` synonym.
+The selections are independent, at least one is required, and each offers `core`, `browser`,
+and `server`.
 
 | Selection   | What it owns                                                             |
 | ----------- | ------------------------------------------------------------------------ |
@@ -98,10 +88,9 @@ includes app/core so the shared transport contracts have one host-independent ow
 8. **Document and prove parity** for every app export and behavioral method: guide,
    examples, manifest index, and the parity specifiers walking the existing `src` and
    `app` roots and every selected alias.
-9. **Verify.** Run the rules' cleanup sweeps over source and tests. Then run the
-   adversarial pass, a mechanical conformance pass, and the repository gates in their
-   required order. Generated CI runs those gates on the declared minimum Node release and
-   on the current major.
+9. **Verify.** Run the rules' cleanup sweeps over source and tests. Then run the review
+   `.agents/orchestration.md` § Size gate names for the change and the repository gates in
+   their required order.
 
 Do not add showcase, authentication, persistence, proxy, styling-system, or product
 policy unless the request requires it. Do not leave placeholders, compatibility shims,

package/dist/host/agents/skills/{orkestrel-build-application → orkestrel-build}/agents/openai.yaml

@@ -1,4 +1,4 @@
 interface:
   display_name: 'Build Orkestrel Application'
   short_description: 'Design and harden Orkestrel app environments'
-  default_prompt: 'Use $orkestrel-build-application to build or harden this workspace application layer.'
+  default_prompt: 'Use $orkestrel-build to build or harden this workspace application layer.'

package/dist/host/agents/skills/orkestrel-debrief/SKILL.md

@@ -5,24 +5,16 @@ description: Look back at a long campaign to learn from its mistakes and success
 
 # Debrief a closed campaign
 
-## Load authority
+## Read
 
-Read the current files in this order:
+`AGENTS.md` § Authority and loading names the files every unit reads. This skill adds the
+references the round needs: [instruction-audit.md](references/instruction-audit.md) before
+auditing the agent, rule, skill, and process layer; [field-testing.md](references/field-testing.md)
+before running or judging a live field pass of an agent-facing surface;
+[retention.md](references/retention.md) before retiring the campaign folder.
 
-1. `AGENTS.md`.
-2. Every applicable `.claude/rules/*.md`; the documentation and quality laws bind every
-   finding this round records and every refinement it produces.
-3. The references this round needs: [instruction-audit.md](references/instruction-audit.md)
-   before auditing the agent/rule/skill/process layer;
-   [field-testing.md](references/field-testing.md) before running or judging a live field
-   pass of an agent-facing surface; [retention.md](references/retention.md) before
-   retiring the campaign folder.
-4. `guides/README.md`, the governing guides for what the campaign built, and `ROADMAP.md` when
-   the repository keeps one.
-
-The user's current instruction wins. The debrief judges the artifact and the process that
-produced it; neither is exempt. Capture successes as deliberately as mistakes, and codify
-a practice that worked so it repeats.
+The debrief judges the artifact and the process that produced it; neither is exempt. Record a
+practice that worked as a refinement, so it repeats.
 
 ## The debrief laws
 

package/dist/host/agents/skills/orkestrel-debrief/references/instruction-audit.md

@@ -6,9 +6,9 @@ evidence-first treatment as any surface.
 
 ## Blind passes, one brief
 
-Run the subjective lane and the objective lane on the SAME brief, in parallel, neither
-seeing the other's answer before both return. `.agents/orchestration.md` § Engine
-assignment decides which engine and which role holds each lane, including under a dark
+Run the subjective lane and the objective lane on the SAME brief, each in a clean context,
+neither seeing the other's answer before reconciliation. `.agents/orchestration.md` § Roles
+and § Benches decide which engine and which role holds each lane, including under a dark
 bench; read the assignment there and name it in the dispatch.
 
 Each lane returns numbered findings, most severe first, and exactly one terminal line:

package/dist/host/agents/skills/orkestrel-debrief/references/retention.md

@@ -6,7 +6,7 @@ keeps it recoverable.
 
 A debrief arrives at its Dispose step, after every finding has a
 carrier. A campaign that accepts with no debrief arrives at acceptance, per
-`.agents/orchestration.md` § Where campaign artifacts live. The procedure is the same through
+`.agents/orchestration.md` § Cleanup. The procedure is the same through
 either door.
 
 Call the `.orkestrel/` folder the campaign folder, never a ledger. The word `ledger` names the
@@ -33,15 +33,15 @@ step.
 
 Rule on every location in this table before deleting anything.
 
-| Location                | What it holds                                                                                                                                              | What the prune does                                                                                                |
-| ----------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------ |
-| `.orkestrel/<package>/` | The campaign folder for a campaign about one package: each unit's brief and report, the audit verdicts, the executed instruments, the acceptance evidence. | Deleted in the prune commit.                                                                                       |
-| `.orkestrel/campaign/`  | The shared campaign folder for a campaign spanning several packages: the wave's plan, its routing ledger, its verdicts.                                    | Deleted in the prune commit.                                                                                       |
-| `tmp/units/`            | A native unit's `<unit>-brief.md` and `<unit>-report.md` pair.                                                                                             | Swept. The durable copy already sits in the campaign folder, so a pair missing there blocks the sweep.             |
-| `tmp/<bench>/`          | A bench unit's brief, report, event stream, final answer, and any login log.                                                                               | Swept after the final gate evidence is recorded. Never committed.                                                  |
-| `tmp/probe/`            | The runtime probes the campaign wrote.                                                                                                                     | Swept. A probe that settled a claim becomes a test before the prune, per `.claude/rules/quality.md` § Instruments. |
-| `ROADMAP.md`            | The repository's sequenced plan of record, where it keeps one.                                                                                             | Kept. Strike the chunks the campaign closed and add the forward work it revealed.                                  |
-| `PROPOSAL.md`           | A proposal for work nobody has ruled on yet.                                                                                                               | Deleted after the work lands or the proposal is refused. The ruling goes in the prune commit message.              |
+| Location                | What it holds                                                                                                                                       | What the prune does                                                                                                                 |
+| ----------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------- |
+| `.orkestrel/<package>/` | One package's live campaign per `.agents/orchestration.md` § .orkestrel layout: its plan, routing ledger, open-seam verdicts, and exposure records. | Deleted in the prune commit.                                                                                                        |
+| `.orkestrel/*.md`       | The ecosystem's records per the same section: the fleet campaign's `plan.md` and `ledger.md`, and the release wave's `release.md`.                  | Deleted in the prune commit of the wave or fleet campaign each records.                                                             |
+| `tmp/units/`            | A native unit's `<unit>-brief.md` and `<unit>-report.md` pair.                                                                                      | Swept after the unit's measurements and unresolved findings are carried into the acceptance record, the campaign ledger, or a test. |
+| `tmp/<bench>/`          | A bench unit's brief, report, event stream, final answer, and any login log.                                                                        | Swept after the final gate evidence is recorded. Never committed.                                                                   |
+| `tmp/probes/`           | The runtime probes the campaign wrote.                                                                                                              | Swept. A probe that settled a claim becomes a test before the prune, per `.claude/rules/quality.md` § Instruments.                  |
+| `ROADMAP.md`            | The repository's sequenced plan of record, where it keeps one.                                                                                      | Kept. Strike the chunks the campaign closed and add the forward work it revealed.                                                   |
+| `PROPOSAL.md`           | A proposal for work nobody has ruled on yet.                                                                                                        | Deleted after the work lands or the proposal is refused. The ruling goes in the prune commit message.                               |
 
 A plan-of-record file is not campaign residue. `ROADMAP.md` outlives every prune and takes the
 campaign's forward work; `PROPOSAL.md` is spent the moment its proposal is ruled on, and a spent
@@ -53,11 +53,11 @@ Sweep the whole `tmp/` tree, not the set of folders this campaign created. A jou
 brief left by an earlier session sits in the same directory, carries no date a reader checks, and
 is read by the next campaign as its own.
 
-- Name each file you find before deleting it. A file you cannot attribute to a closed campaign is
+- Start from `node .agents/skills/orkestrel-dispatch/scripts/sweep.ts --report`, then name each file you find before deleting it. A file you cannot attribute to a closed campaign is
   an open item for the carry check rather than residue.
 - Never sweep while a unit is live. `tmp/units/` and `tmp/<bench>/` hold the briefs and journals
-  live lanes are reading, and the Orchestrator's own instruments belong in its scratchpad for this
-  reason, per `.agents/orchestration.md` § Writing concurrency.
+  live lanes are reading, and the Orchestrator's own instruments live in the checkout's `tmp/` for this
+  reason, per `.agents/orchestration.md` § Permission floor.
 - Delete a probe from the source tree before its unit returns, per `.claude/rules/tests.md`. A
   leaked type probe is a placement-sweep failure rather than a retention question.
 

package/dist/host/agents/skills/orkestrel-dispatch/SKILL.md

@@ -0,0 +1,61 @@
+---
+name: orkestrel-dispatch
+description: >-
+  Dispatch one unit of work to a native subagent or an external bench and bring its result back: write the brief, probe the bench, launch under a cap, watch liveness, read the report, check its citations, accept, and delete the launch copies. Use when `.agents/orchestration.md` routes work to a role, when a bench lane (Grok, Astra, or Opus through the Claude CLI) is needed, or when a command will outlive the turn that starts it.
+---
+
+# Dispatch
+
+`.agents/orchestration.md` decides who runs what and binds the permission floor; this skill runs the dispatch. Read `references/launch.md` before any command that outlives the turn, and `references/bench.md` before any bench lane. Run every step a script performs as that script, from the checkout root, with `node`.
+
+## Scripts
+
+| Script               | Does                                                                                                                                                                                                                                                                                                                                                                                                                 |
+| -------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| `scripts/bench.ts`   | `--cursor [--resolve]`, `--codex [--sandbox MODE]`, or `--claude`, each with `[--model ID] [--cap SECONDS]`: resolves the bench entry, reads its version and login, runs one bounded round trip into `tmp/<bench>/bench.jsonl`, and prints JSON with `live`; `--resolve` prints the Cursor entry alone. Exit 3 when the bench is dark.                                                                               |
+| `scripts/brief.ts`   | `--unit UNIT --lane units\|cursor\|codex\|claude [--subject TEXT]` writes the brief from `.agents/templates/brief.md` and never overwrites a launched one; `--check PATH` lists every slash-bearing path the brief names that does not resolve from the root. Exit 3 when one is missing.                                                                                                                            |
+| `scripts/launch.ts`  | `--journal PATH --errors PATH --cap SECONDS [--status] -- COMMAND [ARGS...]` runs the command with no shell and closed stdin, kills it at the cap (its process tree on Windows), records `git status --porcelain` before and after with `--status`, prints a spawn line carrying the `pid` when the command starts, writes that pid to `<journal>.pid`, and ends the errors file with `exit=`. Exit 124 when capped. |
+| `scripts/login.ts`   | `--codex [--cap SECONDS]` reads `codex login status`, runs `codex login --device-auth` journaled to `tmp/codex/login.log` when it is not logged in, prints the URL and the one-time code as they appear, and polls the status until it answers or the cap ends; `--claude` reads `claude auth status`. Exit 0 logged in, 3 not.                                                                                      |
+| `scripts/result.ts`  | `--cursor\|--claude\|--codex JOURNAL [--out PATH]` prints the session id, the answer length, the run's `exit`, `capped`, and `durationMs` from the errors file, and the journal's `bytes`, `ageMs`, and `lastEvent`. Exit 3 when the journal carries no session or answer.                                                                                                                                           |
+| `scripts/cite.ts`    | `PATH [--root DIR] [--json]` lists every `file:line` citation in a report whose file does not exist or whose line is past the end. Exit 3 when one does not resolve.                                                                                                                                                                                                                                                 |
+| `scripts/helpers.ts` | What every skill script shares: `readOption`, `readOptions`, `readList`, `listFiles`, `readJsonObject`, `readString`, `readNumber`, `normalizeSlashes`, `resolveNpm`, `runNpm`. It runs nothing; import it with the `.ts` extension.                                                                                                                                                                                 |
+| `scripts/sweep.ts`   | `--report` lists leftovers under `tmp/` and `.orkestrel/` by age and deletes nothing; `--tmp [--older-than MINUTES]` deletes stale launch files and refuses while a journal is changing; `--unit UNIT [--delete]` lists or deletes one accepted unit's brief, report, claims, journal, errors, status, pid, and last-message files.                                                                                  |
+
+## Write the brief
+
+1. A small native unit's brief is the dispatch message: scope, law references, expected result. Otherwise run `node .agents/skills/orkestrel-dispatch/scripts/brief.ts --unit <unit> --lane <lane>` and fill every section; write `none` in an empty one. A re-run takes the `-<n>` file the script writes; never edit a launched brief.
+2. Back every fact the unit will act on with the command and output that established it. Check a fact against the tree, not against another artifact.
+3. Scope by what the change makes false: run the `orkestrel-scout` skill's `map.ts --refs` and `--census` for an existing member of any enumerated population the unit extends, and own every file that comes back, with the tests that pin the behavior, the fixtures derived from a constant, and the materialized copy of a template.
+4. Order acceptance criteria cheap-first (regenerate, typecheck, lint, scoped test). Name a timing-sensitive or tree-wide gate as an observation, not a criterion.
+5. State that the executor performs the work itself and spawns nothing. A bridge driver carries the brief unaltered and returns the journal path and session id.
+6. Name the host conditions the unit will hit: shell, sandbox, network, files expected dirty, commands known to fail.
+7. Run `brief.ts --check <brief>` and fix every missing path before the launch.
+
+## Launch
+
+- Native, Claude Code: the Agent tool with the role as `subagent_type`; a Workflow for a fan-out or pipeline. Native, Codex: the custom agent by name.
+- Bench: run `bench.ts --<bench>` once per session before the first lane, then launch the transport contract's command through `launch.ts` in the background with `--status`.
+- Size the cap from the `durationMs` that `result.ts` reported for comparable runs, plus gate time, plus slack. Never let the unit name it.
+- Confirm the launch within a minute with `result.ts`: `bytes` past the header and a small `ageMs`. An instantly dead journal is a failed launch; read its errors file.
+
+## Watch
+
+- Read liveness from the artifact the work produces: owned-file mtimes, `result.ts` on the journal, the counts a suite reports. A wrapper's silence proves nothing.
+- Prove a run dead before relaunching, by the process id in the spawn line `launch.ts` prints at start and in the `<journal>.pid` file it writes, per `references/launch.md` § Kill and relaunch. Never `pgrep -f` or `pkill -f` a pattern from a shell whose own command line contains it.
+- A stalled or cap-killed run: resume with the session id from `result.ts`, or relaunch with a successor brief. Triage per `.agents/orchestration.md` § Deviation protocol.
+
+## Accept
+
+1. Capture a native report to `tmp/units/<unit>-report.md`, or read a bench answer with `result.ts --<bench> <journal> --out <path>`. Refuse a bench report with no journal path and session id.
+2. Run `cite.ts <report>` and discard every claim whose citation does not resolve before reading the rest.
+3. Verify with direct evidence: `git status --porcelain`, the diff, and the scoped validation the brief named. A writer's self-report is not evidence.
+4. Run the review the size gate names, then integrate: apply shared-file patches serially; route an added type, mechanism, or criterion to a successor unit.
+5. Write the acceptance commit message with what changed and what the unit measured; record `durationMs` when it informs a cap or a performance claim.
+6. Run `sweep.ts --unit <unit> --delete` to remove the unit's brief, report, claims, journal, errors, status, pid, and last-message files. Delete any probe the unit left in the tree. In a campaign, update `.orkestrel/<package>/plan.md` and the ledger, and delete a closed seam's verdict. Run `sweep.ts --report` at acceptance and `sweep.ts --tmp [--older-than MINUTES]` after the last live unit exits.
+
+## Refuse
+
+- A brief that contradicts a measurement the campaign already recorded: reconcile first.
+- A unit that asks for an unowned edit: stop it and re-scope.
+- A concurrent writer in the same checkout: queue it.
+- A tree-wide mutating command from a unit: strike it from the brief.

package/dist/host/agents/skills/orkestrel-dispatch/agents/openai.yaml

@@ -0,0 +1,4 @@
+interface:
+  display_name: 'Orkestrel Dispatch'
+  short_description: 'Dispatch one unit to a subagent or a bench and bring its result back'
+  default_prompt: 'Use $orkestrel-dispatch to brief, launch, watch, accept, and sweep one unit of work.'

package/dist/host/agents/skills/orkestrel-dispatch/references/bench.md

@@ -0,0 +1,25 @@
+# Bench lanes
+
+An external engine widens capacity and inherits no authority. Treat every bench output as a proposal until the Orchestrator verifies it against the tree. `.agents/transports/<provider>.md` owns each bench's command, sandbox, and recovery; this file owns what is true of every bench.
+
+## Before the lane
+
+- Probe the bench once per session with `node .agents/skills/orkestrel-dispatch/scripts/bench.ts --cursor`, `--codex`, or `--claude` before its first lane, and again at any dispatch after a failure. `live: false` records the bench dark. A version string or a login status is not liveness.
+- Write the brief with `scripts/brief.ts --unit <unit> --lane <bench>`; briefs never travel as shell arguments. Run `scripts/brief.ts --check` on it and fix every missing path.
+- Name in the brief the host restrictions the transport contract records that this unit will meet. A proof that needs a restricted capability is taken on the host by the Orchestrator, and the unit records it as an observation naming the exact command.
+- When the sandbox rejects a write, the unit stops and reports the rejection. It never tries another write mechanism, and the brief says so.
+- A bench unit writes only its owned files and its checkout's `tmp/`. Name its report path there, or take the report from the last-message file.
+- Launch one Grok lane at a time. An empty lane beside a live probe is starvation, so cut concurrency and re-run rather than recording the bench dark.
+
+## During the lane
+
+- The journal under `tmp/<bench>/` is the liveness signal and the recovery handle; `scripts/result.ts` reads both. Register the lane in the session task registry with subject, journal path, and session id.
+- Never poll the bench from inside the driver. The Orchestrator owns the cap and the watch.
+
+## After the lane
+
+- Refuse a result whose journal path and session id are absent: it ran on the driver's engine.
+- Read the answer with `scripts/result.ts`, never from stdout. Run `scripts/cite.ts` over it before reading it.
+- Verify a writing lane with `git status --porcelain`, the diff, and scoped validation before integrating.
+- A dispatch that fails on auth, quota, model access, or network is a fresh liveness result: record the bench dark and re-plan the lane on the substitute engine.
+- Delete the brief, journal, `.err`, and last-message files when the unit is accepted.

package/dist/host/agents/skills/orkestrel-dispatch/references/launch.md

@@ -0,0 +1,32 @@
+# Long-running commands
+
+Apply these rules to every command that outlives the turn that starts it: a bench exec, a Workflow, an install, a build, a publish chain.
+
+## Launch
+
+- Launch as a harness-tracked background command through `scripts/launch.ts` under a hard cap. Never detach a run from inside a dispatched agent; an unowned run has no completion signal and no death notice.
+- Write a multi-step chain as a TypeScript file run by `node`, with an opening comment naming its log, its cap, and what changed from the file it supersedes. Never edit a script while it runs: copy, edit the copy, launch the copy.
+- On Windows, keep every program-carrying command in a file: a heredoc, `node -e`, an `&&` chain, or a `${...}` argument trips the approval classifier.
+- Run the first use of any flag, subcommand, or quoting form in a throwaway probe, never inside a unit.
+- Keep network-dependent work (installs, lockfile generation, live fetches) out of sandboxed bench execs.
+- Confirm the journal advanced past its header before recording the launch as started; `scripts/result.ts` reports `bytes` and `ageMs`.
+- Never push a git commit through a hosting REST API when a file's content must ride in the request; the per-call cap truncates it.
+
+## Watch
+
+- Arm one Monitor per long exec, filtered to milestones and terminal states, exiting on the terminal event. Never stream the raw journal.
+- A Workflow journal writes only at agent start and result; read liveness from the newest subagent transcript. Recover with `resumeFromRunId`.
+- Judge a unit by the tree it changes: owned-file mtimes, suite counts, its report file. Check before killing anything; a unit killed on a false signal loses unwritten work.
+
+## Kill and relaunch
+
+- Take the child's `pid` from the spawn line `scripts/launch.ts` prints when the command starts, or from `<journal>.pid`. Never take it from a pattern match over command lines, and never run `pgrep -f` or `pkill -f`.
+- On a POSIX host, read liveness with `kill -0 <pid>` (exit 0 means alive), list descendants by running `ps -A -o pid= -o ppid=` and taking the rows whose ppid is the recorded pid, then each listed id in turn, print the list, then kill each listed id and the recorded pid with `kill -KILL <id>`.
+- On Windows under Git Bash, write every Windows slash option as `//`. Read liveness from the output of `tasklist //FI "PID eq <pid>" //NH`, never from its exit code: a row naming `<pid>` means alive, and the `INFO:` line means gone. Alternatively, read the `WINPID` column of `ps -W`. Never read liveness with `kill -0` or `ps -eo`. List descendants with `powershell -NoProfile -Command "Get-CimInstance Win32_Process -Filter 'ParentProcessId=<pid>' | Select-Object -ExpandProperty ProcessId"` applied to each listed id in turn, print the list, then kill the recorded tree with `taskkill //pid <pid> //T //F`.
+- Refuse a list holding PID 1, the harness shell, or an id outside the descent of the recorded pid. At the cap, `launch.ts` kills the tree on Windows and the command alone on any other host; on a POSIX host, kill the listed descendants of a capped run by hand.
+- Confirm the recorded process and every listed descendant are gone before another writer takes the files. A killed `codex exec` is dead only when its `codex-code-mode-host` child is gone too. Check owned-file mtimes before dispatching the substitute.
+- Re-run a timing failure alone, after the unit exits, before believing it; a unit's own re-run is never alone.
+
+## Cap
+
+Size the cap from the `durationMs` that `scripts/result.ts` reported for comparable runs, plus a gate allowance, plus slack. Record the duration beside the result so the next cap has a measurement. A cap-killed exec (`capped=true`, exit 124) reads exactly like a real failure; check the cap first.