@orkestrel/scaffold 0.0.125 → 0.0.127

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (35) hide show
  1. package/dist/bin/main.js +50 -14
  2. package/dist/bin/main.js.map +1 -1
  3. package/dist/host/AGENTS.md +1 -1
  4. package/dist/host/agents/orchestration.md +5 -4
  5. package/dist/host/agents/skills/orkestrel-falsify/SKILL.md +1 -1
  6. package/dist/host/agents/skills/orkestrel-harden/references/methods.md +2 -2
  7. package/dist/host/agents/skills/orkestrel-journey/SKILL.md +18 -19
  8. package/dist/host/agents/skills/orkestrel-parallel/references/codex.md +2 -1
  9. package/dist/host/claude/AGENTS.md +2 -2
  10. package/dist/host/claude/agents/orkestrel.md +18 -18
  11. package/dist/host/claude/agents/planner.md +2 -2
  12. package/dist/host/claude/rules/quality.md +3 -0
  13. package/dist/host/claude/rules/tests.md +3 -0
  14. package/dist/host/claude/rules/workspace.md +11 -5
  15. package/dist/host/codex/agents/planner.toml +2 -2
  16. package/dist/host/guides/agent.md +285 -161
  17. package/dist/host/guides/capture.md +123 -123
  18. package/dist/host/guides/database.md +18 -7
  19. package/dist/host/guides/ndjson.md +4 -3
  20. package/dist/host/guides/ollama.md +2 -2
  21. package/dist/host/guides/program.md +3 -3
  22. package/dist/host/guides/qualifier.md +1 -1
  23. package/dist/host/guides/relation.md +71 -31
  24. package/dist/host/guides/scaffold.md +101 -64
  25. package/dist/host/guides/study.md +22 -12
  26. package/dist/host/guides/toolbox.md +9 -2
  27. package/dist/host/manifest.json +26 -26
  28. package/dist/host/tests/config.test.ts +24 -6
  29. package/dist/src/core/index.cjs +134 -79
  30. package/dist/src/core/index.cjs.map +1 -1
  31. package/dist/src/core/index.d.cts +30 -7
  32. package/dist/src/core/index.d.ts +30 -7
  33. package/dist/src/core/index.js +134 -80
  34. package/dist/src/core/index.js.map +1 -1
  35. package/package.json +4 -4
@@ -64,7 +64,7 @@ configs/ thin target wrappers around root Vite/TypeScript configuration
64
64
  - **Functional core, imperative shell.** Pure exported leaves; stateful orchestration as class methods. A method never forwards 1:1 to a helper.
65
65
  - **No superfluous wrappers.** A wrapper adds a boundary, invariant, composition, translation, lifecycle, or materially narrower contract, or it goes.
66
66
  - **Minimal public API.** Create or substantively expand a capability with its first real consumer; this gate applies at creation, never later. Expose an existing reusable capability through its environment barrel regardless of consumer count. Remove a symbol only when the capability itself must not exist. Prefer one minimal interface and one shared engine, with a native backend override only for a faster path.
67
- - **No compatibility shims.** Update every consumer in the same change.
67
+ - **No compatibility shims.** Update every consumer in the same change. Fleet packages and applications are greenfield: write no alias, wrapper, or migration for a shape or for data an earlier release produced.
68
68
  - **Mechanism, not product policy.** Framework code stops before application decisions.
69
69
  - **No polling architecture.** Park idle work on events and abort signals. Yield long work cooperatively.
70
70
 
@@ -14,7 +14,7 @@ Name the harness family and its access route: Claude through Claude Code or the
14
14
 
15
15
  | Engine | Job | Posture |
16
16
  | ----------------- | ------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
17
- | Claude Fable 5.1 | orchestration in Claude Code: the plan, every decision, integration, acceptance | never dispatched as a writer, a bench, a fork, or a Workflow node; dispatched only as a read-only consultant, for a bounded design question at the stop the Orchestrator set for a seam, or for the wording and coherence pass the user names |
17
+ | Claude Fable 5.2 | orchestration in Claude Code: the plan, every decision, integration, acceptance | never dispatched as a writer, a bench, a fork, or a Workflow node; dispatched only as a read-only consultant, for a bounded design question at the stop the Orchestrator set for a seam, or for the wording and coherence pass the user names |
18
18
  | Cursor Grok | context absorption, distillation, broad mapping, exploration | as a bench: read-only; returns evidence, never decisions |
19
19
  | Claude Opus 5.5 | subjective design, design-fit review, implementation | proposes, audits, implements; never accepts its own work |
20
20
  | GPT-6 Astra | objective analysis, correctness audit, implementation | proposes, audits, implements; never accepts its own work |
@@ -23,7 +23,7 @@ Name the harness family and its access route: Claude through Claude Code or the
23
23
  | GPT-6.1 Sol | a fully specified unit with a closed shape | executes a fully specified brief |
24
24
  | GPT-6 Luna | mechanical units, drivers, small precise tasks, gates, exact lookups | executes a fully specified brief |
25
25
 
26
- The harness's own engine orchestrates: Claude Fable 5.1 or Claude Opus 5.5 in Claude Code, GPT-6 Astra in Codex, Cursor Grok in Cursor. The Orchestrator owns the plan, every decision, integration, and acceptance. When the harness is Codex or the Orchestrator uses a Codex-family engine, use the Codex column: map Opus to GPT-6 Astra, Sonnet to GPT-6.1 Sol, and Haiku to GPT-6 Luna; execute mapped Astra roles natively without an Opus bench probe. Under this mapping, assign review to another agent in a clean context and forbid self-review, including when writer and reviewer both use Astra. For other harnesses, review runs on an engine that did not write the work.
26
+ The harness's own engine orchestrates: Claude Fable 5.2 or Claude Opus 5.5 in Claude Code, GPT-6 Astra in Codex, Cursor Grok in Cursor. The Orchestrator owns the plan, every decision, integration, and acceptance. When the harness is Codex or the Orchestrator uses a Codex-family engine, use the Codex column: map Opus to GPT-6 Astra, Sonnet to GPT-6.1 Sol, and Haiku to GPT-6 Luna; execute mapped Astra roles natively without an Opus bench probe. Under this mapping, assign review to another agent in a clean context and forbid self-review, including when writer and reviewer both use Astra. For other harnesses, review runs on an engine that did not write the work.
27
27
 
28
28
  ## Roles
29
29
 
@@ -48,7 +48,7 @@ The harness's own engine orchestrates: Claude Fable 5.1 or Claude Opus 5.5 in Cl
48
48
  - A role holding the subjective lane judges whether the requested shape and voice are present, names and boundaries are coherent, the work sits at the right layer, each concept earns its place, and the guide matches the code. A role holding the objective lane judges correctness under adverse orderings, what the contracts and installed declarations permit, dependency and range truth, the missing seam or the assertion that cannot fail, and the letter of the rules. A charter points here for both lenses.
49
49
  - In a falsify round, where a claim about a name or a comment is no claim, both lanes hold the same claims and differ by posture: the subjective lane drives the subject as a real caller would and reports what that caller meets, and the objective lane tests the contract's letter across adverse orderings.
50
50
  - Read the word `lane` by its qualifier: a subjective or an objective lane is the lens a role holds, a writer, runner, or review lane is a unit at work, and the execution lane is the host resource `.agents/skills/orkestrel-parallel/SKILL.md` § Share an execution lane between sessions orders. Read the word `census` the same way: the `map.ts --census` flag counts a term, the `discovery.ts` script lists test collection, `.agents/skills/orkestrel-harden/references/bindings.md` sweeps a binding key, and a class census under § Depth, breadth, and machinery lists one defect class across packages.
51
- - `planner` also carries the read-only consultation § Engines admits for Claude Fable 5.1, on the engine the dispatch names.
51
+ - `planner` also carries the read-only consultation § Engines admits for Claude Fable 5.2, on the engine the dispatch names.
52
52
  - On a harness with no column, map each capability to an engine in the roster and route by that map; add the harness's column here when it carries campaign work.
53
53
  - A driver launches another harness's CLI and returns the journal path and session id with the result. It never judges or implements. Refuse a bench result that carries no journal path and session id; it ran on the driver.
54
54
  - `.agents/transports/<harness>.md` owns each bench's invocation, sandbox, journal, and recovery. `.claude/agents/` and `.codex/agents/` hold one file per role.
@@ -76,6 +76,7 @@ The harness's own engine orchestrates: Claude Fable 5.1 or Claude Opus 5.5 in Cl
76
76
  - On continuation, the Orchestrator sends the next bounded objective and the facts changed after the last turn. The continued agent reads changed or missing authority and the source regions that objective needs, and rechecks mutable facts such as source state and process liveness. Do not repeat unchanged bulk reads merely because a unit resumes. A continuation that changes scope or acceptance, or authorizes a fix, is written first as `<unit>-brief-<n>.md` and sent as a pointer to that file; a message alone continues only an unchanged objective.
77
77
  - Consult a suitable peer at the stop set for a seam, or earlier when a consequential approach remains open, evidence conflicts, or attempts stop narrowing a failure, without waiting for the user, and record the question, recommendation, evidence, and next check in the plan. Give the peer the question, relevant evidence, attempted approaches, constraints, and requested answer. Use an existing specialist when its context helps and a fresh peer when the question needs an independent challenge.
78
78
  - Keep consultation read-only and bounded. Ask for a recommendation, its evidence, uncertainty, and a concrete next check. The executor retains the work; the Orchestrator rules on scope and acceptance. Advice grants no authority and does not satisfy an independent review gate.
79
+ - Before a writer builds a consultant's proposal, test each claim the proposal rests on with probes on both arms, the base and the proposal applied by temporary mutation, and rule from those readings, never from the consultant's reasoning.
79
80
  - Resolve conflicting advice against the contract and observable evidence. Do not add consultation to routine steps or repeat it without a changed question or evidence.
80
81
  - When a requested launch channel fails, record the failure and whether a child was created. Before retrying an uncertain launch, inspect the recorded task or session. Use a permitted fallback with its actual model and options disclosed; retry a refused route only after a relevant condition changes.
81
82
 
@@ -134,7 +135,7 @@ The harness's own engine orchestrates: Claude Fable 5.1 or Claude Opus 5.5 in Cl
134
135
  - A review judges types, tests, and implementation. It reports comment, TSDoc, or guide wording only where the text contradicts the final behavior, and no fix round opens for prose alone.
135
136
  - When a brief restates a ruling, give the reviewer the ruling's own text beside the brief, and make any disagreement between the two a claim.
136
137
  - Before a repair is dispatched, class its remedy as a correction, a design question, or a deferral; a remedy that adds a mechanism (a file operation, a state, a protocol step, or a retry class) is a design question that goes to one `planner` or `analyst` pass carrying measured platform facts.
137
- - A fix round re-audits the repaired claim only. A fix that departs from the reviewer's prescription gets one pass by an independent reviewer per § Engines.
138
+ - A fix round re-audits the repaired claim only, in a fresh context. After the re-audit's findings are repaired, one closure check reads whether each closed and tests no other claim. A fix that departs from the reviewer's prescription gets one pass by an independent reviewer per § Engines.
138
139
  - Every repair, a prescription adopted verbatim included, closes with a mutation probe on its load-bearing line (disable it, watch the test fail, restore it); `node .agents/skills/orkestrel-falsify/scripts/mutate.ts --edits EDITS.md -- COMMAND` runs the probe and proves the restore. Where the repair adds or changes a process proof, the proof's repeat run under suite load closes it too.
139
140
  - Before another round at the same seam, weigh whether a further repair beats a design question, set and state in the brief the stop for that seam and its reason, re-examine it at every round per § Depth, breadth, and machinery, and end the depth search with a design question when a round introduces as many defects as it closes, a repair adds a mechanism, findings stop converging, or the cost outgrows the goal's importance. `.agents/skills/orkestrel-falsify/SKILL.md` § End the depth search owns how the search ends.
140
141
  - A repair round at a seam carries only rows whose remedy serves that seam; a finding outside it waits for the next round or the deferral list.
@@ -12,7 +12,7 @@ One round. It ends with a ruling. `.agents/orchestration.md` § Size gate decide
12
12
 
13
13
  Run from the checkout root. The script's opening comment owns its behavior.
14
14
 
15
- - `scripts/mutate.ts --edits EDITS.md -- COMMAND [ARGS...]` runs the mutation probe. It resolves an edit file, the `### <path>` block shape that `.agents/skills/orkestrel-dispatch/scripts/patch.ts` reads, before it touches a file; saves the originals under `tmp/probes/mutate/`; writes the mutated files; runs the command with no shell and its output on standard error; restores the bytes and proves each file's SHA-256 digest; runs the command again; and prints one JSON line with both exits. A mutation that leaves the suite uncollected also reads as a failing run, so read the command's own test count on standard error. A run killed mid-command leaves the tree mutated: the next run refuses to start while a run directory remains and names the line that restores it, `scripts/mutate.ts --restore tmp/probes/mutate/run-NAME`, which writes the saved originals back, proves each digest from the manifest, and deletes the directory. Exit 0 when the mutated run failed and the restored run passed (or a `--restore` is proven), 3 when either did not, 2 when a restore is not proven or the command cannot start, 64 usage, an edit that does not resolve, or a run directory left behind.
15
+ - `scripts/mutate.ts --edits EDITS.md -- COMMAND [ARGS...]` runs the mutation probe. It resolves an edit file, the `### <path>` block shape that `.agents/skills/orkestrel-dispatch/scripts/patch.ts` reads, before it touches a file; saves the originals under `tmp/probes/mutate/`; writes the mutated files; runs the command with no shell and its output on standard error; restores the bytes and proves each file's SHA-256 digest; runs the command again; and prints one JSON line with both exits. A mutation that leaves the suite uncollected also reads as a failing run, so read the command's own test count on standard error. When the probe compares two arms rather than proving a kill, read the `mutatedExit` and `restoredExit` fields from the JSON line: exit 3 reports a passing mutated run. A run killed mid-command leaves the tree mutated: the next run refuses to start while a run directory remains and names the line that restores it, `scripts/mutate.ts --restore tmp/probes/mutate/run-NAME`, which writes the saved originals back, proves each digest from the manifest, and deletes the directory. Exit 0 when the mutated run failed and the restored run passed (or a `--restore` is proven), 3 when either did not, 2 when a restore is not proven or the command cannot start, 64 usage, an edit that does not resolve, or a run directory left behind.
16
16
 
17
17
  ## Run a round the size gate does not name when
18
18
 
@@ -249,7 +249,7 @@ When a comparison has a written pairwise graph, select by that graph, preserve p
249
249
 
250
250
  ### Compare the shippable intervention, including what it loses
251
251
 
252
- When copying, quotes, aliases, units, field kinds, or reader size is proposed as a treatment, evaluate each as a different intervention, run with only metadata the public contract carries, set the keep rule as a cost ratio before the column runs, retain latent errors and lost right answers, and inspect shared brief vocabulary in purported holdouts.
252
+ When copying, quotes, aliases, units, field kinds, or reader size is proposed as a treatment, evaluate each as a different intervention, run with only metadata the public contract carries, set the keep rule as a cost ratio before the column runs, counting a refusal's removed right values apart from its removed wrong values, retain latent errors and lost right answers, compare each arm with its rival on the same rows at the same coverage with a minimum coverage per class, give the rival every input the arm sees, and inspect shared brief vocabulary in purported holdouts.
253
253
 
254
254
  - **Refuse.** Usage frequency as counterfactual benefit, because a usage count measures wording, not value. An additive oracle as the replacement fitted method, because the two are different interventions. Field descriptions fitted from the writers' own brief, or fitted synonyms, called independent generalization, because writers who share descriptions share vocabulary, so quote only blind readings. Filling fields by a model's index links or quoted evidence, because index links add silent errors and quoting costs more right answers than the wrong answers it removes. Unit lists through alias channels, because units collide. A provider's need for prompt teaching of structured-output descriptions generalized to every provider, because that behavior is recorded on one provider.
255
255
  - **Reading.** In the `@orkestrel/learn` package, on the reader research quote-field intervention: quote fields remove six accepted wrongs but cost 99 right answers and raise latent errors.
@@ -471,7 +471,7 @@ When an expensive use depends on a freeze, a grade, progress, a replay, or next-
471
471
 
472
472
  ### Separate developmental evidence, confirmation, and answer-path quality
473
473
 
474
- When a study reports a measurement, hold the populations apart as the law states, fit preprocessing only on training data, measure accepted wrong, latent wrong, useful coverage, and the answers-only user's final decision, run helpful correction as a separate scenario, keep unknown, unselected, and unreached distinct, and replace a changed subject's assertions.
474
+ When a study reports a measurement, hold the populations apart as the law states, fit preprocessing only on training data, measure accepted wrong, latent wrong, useful coverage, and the answers-only user's final decision, read the rows that decide a change by hand beside its score, run helpful correction as a separate scenario, keep unknown, unselected, and unreached distinct, and replace a changed subject's assertions.
475
475
 
476
476
  - **Refuse.** An old green result carried to a changed subject, because “none survives as a proof”. A labeling floor as a quality control, because that floor “cannot fail”. A label disagreement on equal products as a product difference, because that disagreement “is the labelers' own”. Semantic substitution from a lexical score, because its correlations did not earn that substitution. A population audit that asserts counts alone, because every count can hold while strata fail to cross, one value dominates, boundary values land together, or a ceiling under the bar makes the run measure the grid: in `@orkestrel/learn`, every count holds while all 200 ordinary messages are approvals and one pair of boundary values sits on 202 of 600 subjects. An audit with no negative control, because an audit proves itself by refusing the draw it was written to reject, so rerun it on the prior draw, per `.claude/rules/quality.md` § Instruments. A sum of sole-blocker rows as a bound, because removing one blocker changes which rows the others block and can add wrong rows: report each removal's row-level right and wrong, as `@orkestrel/learn` read `Superseded` gaining 126 right and adding 0 wrong and `Support` gaining 5 right and adding 1 wrong.
477
477
  - **Reading.** In the `@orkestrel/learn` package, on its study: “only the rubric contains population membership and expected outputs”. The answers-path scenario stays a specification, not a completed result, and no completed carrier is recorded.
@@ -64,11 +64,11 @@ Resolve every target and every population by ARIA role and accessible name as re
64
64
 
65
65
  ## Read the variant once
66
66
 
67
- A browser application born by `scaffold new` carries the journey axis: the
68
- `configs/app/vite.journey.config.ts` wrapper, the root `appJourney` factory, the `test:journey`
69
- script, and that script's place in the `test` chain. That wiring fans the journey suite out into one
70
- Vitest project per variant, and each project provides its own variant. Read the variant once at run
71
- start, and let it choose the capture destination, the matrix row, and the statechart run together.
67
+ For a browser application born by `scaffold new`, use its journey axis: the
68
+ `configs/app/vite.journey.config.ts` wrapper, the root `appJourney` factory, and the `test:journey`
69
+ script. Place the script by `.claude/rules/workspace.md` § Test project matrix.
70
+ Read the variant once at run start from the Vitest project the wrapper creates for it, and let it
71
+ choose the capture destination, the matrix row, and the statechart run together.
72
72
 
73
73
  - Read `inject('variant')` for this run's variant name, `inject('variants')` for the declared list,
74
74
  and `inject('capture')` for whether this run writes frames.
@@ -79,16 +79,14 @@ start, and let it choose the capture destination, the matrix row, and the statec
79
79
  viewports for this application.
80
80
  - Activate the axis in a workspace born before it: write `configs/app/vite.journey.config.ts`, run
81
81
  `scaffold audit`, and add any missing planned dependencies it names before `scaffold repair`.
82
- Then add `npm run test:journey` to the `test` script after `npm run test:app`.
83
- The repair defines `appJourney` in the root configuration, emits `test:journey`, `journey`, and
84
- `capture`, and excludes the journey suite from the ordinary `app:browser` project. It does not rewrite the `test`
85
- chain, so the chain entry is yours to add. Add that entry before the repair instead where you
86
- prefer: `test:journey` names a configuration rather than a project, so neither order blocks the
87
- repair.
82
+ Check that the repair defines `appJourney` in the root configuration, emits `test:journey`,
83
+ `journey`, and `capture`, and excludes the journey suite from the ordinary `app:browser` project.
84
+ Place each script by `.claude/rules/workspace.md` § Test project matrix. When the repair preserves
85
+ an authored `test` chain, apply that placement to the chain yourself.
88
86
  - Name each variant per [captures.md](references/captures.md) → Variants.
89
87
  - Keep the provided variants as serializable `JourneyVariant` data: `name`, `width`, and `height`.
90
88
  Compose no callback merely to pass that data to a capture.
91
- - Keep `npm run test:journey` as the whole-suite gate in the `test` chain. Use the scoped commands
89
+ - Use `npm run test:journey` for the whole suite. Use the scoped commands
92
90
  under Choosing what to run for iteration and stamped frames; keep the root configuration's
93
91
  `CAPTURE === '1'` read and its `capture` provide unchanged.
94
92
  - Keep the journeys in `tests/app/browser/integration.test.ts`, and the journeys of the `vue`
@@ -103,12 +101,14 @@ start, and let it choose the capture destination, the matrix row, and the statec
103
101
  `scaffold audit` reports a manifest and a wrapper that disagree as one of these findings. Settle the one it reports before trusting a green run.
104
102
 
105
103
  - The manifest names a Vitest configuration for `test:journey` that the plan does not emit and the target does not hold. Write the wrapper and run `scaffold repair`, or remove the `test:journey` script and its chain invocation.
106
- - The manifest does not invoke `npm run test:journey` from its `test` chain. Add `npm run test:journey` to the `test` script after `npm run test:app`.
104
+ - When a private manifest does not invoke `npm run test:journey` from its `test` chain, add that invocation after `npm run test:app`.
105
+ - When a publishing manifest's `test` chain exactly matches the former generated chain, run `scaffold repair` or `scaffold overwrite` to remove its application and journey gate invocations. When the chain differs, remove those invocations yourself. Run the helper suites on demand.
107
106
 
108
- `scaffold audit` reports the invocation question only when no chain from `test` reaches
109
- `npm run test:journey` through literal `npm run` calls, so a chain reaching it through an
110
- intermediate script raises none. It reports the configuration question only for a `test:*` script
111
- whose text names `vitest`, so a script naming another runner's configuration raises none either.
107
+ For a private workspace's missing-invocation question, inspect literal `npm run` calls through
108
+ intermediate scripts as well as the `test` script. For a publishing helper-suite question, inspect
109
+ the reported caller and invocation, including literal Vitest calls through application and journey wrappers.
110
+ For an absent-configuration question, inspect
111
+ `test:*` scripts whose text names `vitest`.
112
112
 
113
113
  ### Choosing what to run
114
114
 
@@ -152,8 +152,7 @@ Climb this ladder using only declared owner, case, and variant names:
152
152
  before `npm run capture -- --changed HEAD`.
153
153
  3. A section or page: `npm run journey -- pages:mail:inbox` or `npm run journey -- pages:mail`.
154
154
  4. A group: `npm run journey -- pages`.
155
- 5. `all` at the gate: `npm run journey -- all`, retaining `npm run test:journey` in the repository's
156
- whole-suite test chain.
155
+ 5. The whole suite: `npm run journey -- all`.
157
156
 
158
157
  Follow a refusal's exact next commands before widening a selection. Use the matching `capture`
159
158
  script only when the work needs frames, and apply [captures.md](references/captures.md) before
@@ -20,11 +20,12 @@ Native Codex agents, and Codex lanes another harness hosts. `.codex/config.toml`
20
20
 
21
21
  ## Quirks
22
22
 
23
- - Before a Codex lane that requires a tool server, verify the entry's resolution and handshake from that lane's checkout under the effective configuration. On "required MCP servers failed to initialize", read the user-level configuration for a same-named entry.
23
+ - Before a Codex lane that requires a tool server, verify the entry's resolution and handshake from that lane's checkout under the effective configuration. On "required MCP servers failed to initialize", read the user-level configuration for a same-named entry: Codex merges its keys into the project's entry, so a user-level `cwd` resolves the project's relative `args` from that directory (2026-11-08). Remove the user-level `cwd` rather than give the project entry an absolute path.
24
24
  - When a Codex lane runs a long command, brief the yield fallback before launch: the longest yield interval, output to a file, process-list reads until the tracked command exits, and no kill. The lane cannot set a timeout; brief it to proceed without asking for one.
25
25
  - When the shell yields before the command exits, start nothing else, and confirm no owned leftover remains before the next lane; a lane that started the next test file broke the one-runner bound (2026-10-22).
26
26
  - When a brief launches a long command, name its terminal record, such as the launcher's `exit=` line; a lane read a finished step as running and polled it for six minutes (2026-10-22).
27
27
  - When a review brief describes risks in offensive-security wording, the provider's filter can end the lane without a verdict (2026-10-20). Brief the review as quality assurance of the user's own code in neutral words, require an early verdict file, retry once, then use a recorded fallback.
28
+ - When a Codex lane writes probes under `tmp/probes/`, brief it to name its probe folder in its report and to leave the folder when the automatic approval review refuses the deletion (`blocked by policy`); delete the folder after the lane returns.
28
29
  - When a `codex exec` run is killed, treat it as dead only when its `codex-code-mode-host` child is gone too.
29
30
  - When a Codex child runs through a hosting harness, it can reach that harness's messaging tools; probe them before a writer or runner depends on them (`references/t3.md`).
30
31
  - When a launch fails on a tool server's handshake timeout, relaunch that lane alone; a lane beside it started with the same server (2026-10-19). Stagger launches that start the same server.
@@ -10,9 +10,9 @@
10
10
 
11
11
  ## Models
12
12
 
13
- - Use the aliases `haiku`, `sonnet`, and `opus` in role frontmatter, a Workflow node, and a per-invocation `model`. Never `fable`, except the consultant dispatch `.agents/orchestration.md` § Engines admits, which passes `model: fable` per invocation; never `inherit`, never a fixed model ID, and never a dispatch that takes its model from the session, including an `Explore` call or a Workflow node without `model`; `.agents/orchestration.md` § Engines owns where Claude Fable 5.1 may run.
13
+ - Use the aliases `haiku`, `sonnet`, and `opus` in role frontmatter, a Workflow node, and a per-invocation `model`. Never `fable`, except the consultant dispatch `.agents/orchestration.md` § Engines admits, which passes `model: fable` per invocation; never `inherit`, never a fixed model ID, and never a dispatch that takes its model from the session, including an `Explore` call or a Workflow node without `model`; `.agents/orchestration.md` § Engines owns where Claude Fable 5.2 may run.
14
14
  - Keep `CLAUDE_CODE_SUBAGENT_MODEL` at `sonnet` in the `env` object of `.claude/settings.json`. It serves a subagent that no role frontmatter and no dispatch assigned a model; frontmatter and a per-invocation `model` outrank it (measured 2026-10-08 on Claude Code 2.1.294: `general-purpose` took it, `Plan` did not). Never set `CLAUDE_CODE_SUBAGENT_MODEL_FORCE`; it overrides every role.
15
- - An alias serves the model the installed Claude Code maps it to. After a model release, measure the mapping with `claude -p "Reply with the single word OK." --model <alias> --output-format json --max-turns 1` and read `modelUsage`; when a newer model answers by id and the alias lags, update Claude Code through the host's package manager and measure again. Claude Code 2.1.294 serves Haiku 5.5 as `haiku`, Sonnet 5.5 as `sonnet`, and Opus 5.5 as `opus` (measured 2026-10-08). When a measurement moves an alias, update, in the same change, the version each `.claude/agents/` charter names in its description and opening line and the rows of `.agents/orchestration.md` § Engines.
15
+ - An alias serves the model the installed Claude Code maps it to. After a model release, measure the mapping with `claude -p "Reply with the single word OK." --model <alias> --output-format json --max-turns 1` and read `modelUsage`; when a newer model answers by id and the alias lags, update Claude Code through the host's package manager and measure again. Claude Code 2.1.294 serves Haiku 5.5 as `haiku`, Sonnet 5.5 as `sonnet`, and Opus 5.5 as `opus` (measured 2026-10-08). Claude Code 2.1.321 serves Fable 5.2 as `fable`, and 2.1.319 serves Fable 5.1 (measured 2026-11-09); run 2.1.321 or later. When a measurement moves an alias, update, in the same change, the version each `.claude/agents/` charter names in its description and opening line and the rows of `.agents/orchestration.md` § Engines.
16
16
  - Run the main session on `fable` or `opus` at high effort. Apply `.agents/orchestration.md` § Cache; check Claude Code's documented behavior for the installed version, model, and provider before changing model options.
17
17
  - Reach Astra through `analyst` and `astra`, Grok through `grok`. Role frontmatter carries Claude models only.
18
18
 
@@ -44,16 +44,16 @@ Treat generated package identifiers, versions, layers, and dependency ranges as
44
44
  | Package | Version | Layer | Runtime dependencies | Peer dependencies |
45
45
  | ----------------------- | --------- | ----- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------- |
46
46
  | `@orkestrel/abort` | `0.0.16` | L1 | `@orkestrel/contract` `^0.0.22` | |
47
- | `@orkestrel/agent` | `0.0.36` | L5 | `@orkestrel/tool` `^0.0.22`, `@orkestrel/abort` `^0.0.16`, `@orkestrel/queue` `^0.0.20`, `@orkestrel/budget` `^0.0.17`, `@orkestrel/ndjson` `^0.0.16`, `@orkestrel/emitter` `^0.0.15`, `@orkestrel/timeout` `^0.0.16`, `@orkestrel/contract` `^0.0.22`, `@orkestrel/database` `^0.0.21`, `@orkestrel/workflow` `^0.0.26`, `@orkestrel/workspace` `^0.0.15` | |
47
+ | `@orkestrel/agent` | `0.0.39` | L5 | `@orkestrel/tool` `^0.0.22`, `@orkestrel/abort` `^0.0.16`, `@orkestrel/queue` `^0.0.21`, `@orkestrel/budget` `^0.0.17`, `@orkestrel/ndjson` `^0.0.16`, `@orkestrel/emitter` `^0.0.15`, `@orkestrel/timeout` `^0.0.16`, `@orkestrel/contract` `^0.0.22`, `@orkestrel/database` `^0.0.22`, `@orkestrel/workflow` `^0.0.27`, `@orkestrel/workspace` `^0.0.16` | |
48
48
  | `@orkestrel/brief` | `0.0.17` | L4 | `@orkestrel/reason` `^0.0.17`, `@orkestrel/emitter` `^0.0.15`, `@orkestrel/contract` `^0.0.22`, `@orkestrel/interpret` `^0.0.22` | |
49
49
  | `@orkestrel/browser` | `0.0.32` | L5 | `@orkestrel/mcp` `^0.0.42`, `@orkestrel/html` `^0.0.15`, `@orkestrel/pool` `^0.0.20`, `@orkestrel/tool` `^0.0.22`, `@orkestrel/codec` `^0.0.9`, `@orkestrel/router` `^0.0.20`, `@orkestrel/server` `^0.0.26`, `@orkestrel/emitter` `^0.0.15`, `@orkestrel/process` `^0.0.18`, `@orkestrel/contract` `^0.0.22`, `@orkestrel/markdown` `^0.0.20`, `@orkestrel/websocket` `^0.0.18` | |
50
50
  | `@orkestrel/budget` | `0.0.17` | L1 | `@orkestrel/contract` `^0.0.22` | |
51
- | `@orkestrel/capture` | `0.0.3` | L3 | `@orkestrel/process` `^0.0.17`, `@orkestrel/timeout` `^0.0.15`, `@orkestrel/contract` `^0.0.21` | `vite` `^8.3.5`, `vitest` `^4.1.12 \|\| ^5.0.5`, `playwright` `^1.61.0`, `@vitest/browser-playwright` `^4.1.12 \|\| ^5.0.5` |
51
+ | `@orkestrel/capture` | `0.0.5` | L3 | `@orkestrel/process` `^0.0.18`, `@orkestrel/timeout` `^0.0.16`, `@orkestrel/contract` `^0.0.22` | `vite` `^8.3.7`, `vitest` `^4.1.12 \|\| ^5.1.0`, `playwright` `^1.61.0`, `@vitest/browser-playwright` `^4.1.12 \|\| ^5.0.5` |
52
52
  | `@orkestrel/codec` | `0.0.9` | L1 | `@orkestrel/contract` `^0.0.22` | |
53
53
  | `@orkestrel/console` | `0.0.19` | L2 | `@orkestrel/emitter` `^0.0.15`, `@orkestrel/contract` `^0.0.22` | |
54
54
  | `@orkestrel/contract` | `0.0.22` | L0 | | |
55
- | `@orkestrel/csv` | `0.0.12` | L2 | `@orkestrel/codec` `^0.0.8`, `@orkestrel/contract` `^0.0.21` | |
56
- | `@orkestrel/database` | `0.0.21` | L2 | `@orkestrel/sqlite` `^0.0.17`, `@orkestrel/emitter` `^0.0.15`, `@orkestrel/contract` `^0.0.22`, `@orkestrel/indexeddb` `^0.0.17` | |
55
+ | `@orkestrel/csv` | `0.0.13` | L2 | `@orkestrel/codec` `^0.0.9`, `@orkestrel/contract` `^0.0.22` | |
56
+ | `@orkestrel/database` | `0.0.22` | L2 | `@orkestrel/sqlite` `^0.0.17`, `@orkestrel/emitter` `^0.0.15`, `@orkestrel/contract` `^0.0.22`, `@orkestrel/indexeddb` `^0.0.17` | |
57
57
  | `@orkestrel/emitter` | `0.0.15` | L1 | `@orkestrel/contract` `^0.0.22` | |
58
58
  | `@orkestrel/form` | `0.0.16` | L2 | `@orkestrel/emitter` `^0.0.15`, `@orkestrel/contract` `^0.0.22` | |
59
59
  | `@orkestrel/guide` | `0.0.30` | L3 | `@orkestrel/contract` `^0.0.22`, `@orkestrel/markdown` `^0.0.20` | |
@@ -64,37 +64,37 @@ Treat generated package identifiers, versions, layers, and dependency ranges as
64
64
  | `@orkestrel/lsp` | `0.0.14` | L3 | `@orkestrel/emitter` `^0.0.15`, `@orkestrel/jsonrpc` `^0.0.2`, `@orkestrel/process` `^0.0.18`, `@orkestrel/contract` `^0.0.22` | |
65
65
  | `@orkestrel/markdown` | `0.0.20` | L2 | `@orkestrel/html` `^0.0.15`, `@orkestrel/contract` `^0.0.22` | |
66
66
  | `@orkestrel/mcp` | `0.0.42` | L4 | `@orkestrel/sse` `^0.0.13`, `@orkestrel/tool` `^0.0.22`, `@orkestrel/codec` `^0.0.9`, `@orkestrel/emitter` `^0.0.15`, `@orkestrel/jsonrpc` `^0.0.2`, `@orkestrel/process` `^0.0.18`, `@orkestrel/contract` `^0.0.22`, `@orkestrel/websocket` `^0.0.18` | `@orkestrel/router` `^0.0.20`, `@orkestrel/server` `^0.0.26` |
67
- | `@orkestrel/middleware` | `0.0.27` | L4 | `@orkestrel/abort` `^0.0.15`, `@orkestrel/budget` `^0.0.16`, `@orkestrel/timeout` `^0.0.15`, `@orkestrel/contract` `^0.0.21` | `@orkestrel/server` `^0.0.25`, `@orkestrel/database` `^0.0.20` |
68
- | `@orkestrel/msg` | `0.0.15` | L2 | `@orkestrel/codec` `^0.0.8`, `@orkestrel/contract` `^0.0.21` | |
67
+ | `@orkestrel/middleware` | `0.0.29` | L4 | `@orkestrel/abort` `^0.0.16`, `@orkestrel/budget` `^0.0.17`, `@orkestrel/timeout` `^0.0.16`, `@orkestrel/contract` `^0.0.22` | `@orkestrel/server` `^0.0.26`, `@orkestrel/database` `^0.0.22` |
68
+ | `@orkestrel/msg` | `0.0.16` | L2 | `@orkestrel/codec` `^0.0.9`, `@orkestrel/contract` `^0.0.22` | |
69
69
  | `@orkestrel/ndjson` | `0.0.16` | L1 | `@orkestrel/contract` `^0.0.22` | |
70
- | `@orkestrel/ollama` | `0.0.30` | L6 | `@orkestrel/tool` `^0.0.22`, `@orkestrel/agent` `^0.0.36`, `@orkestrel/budget` `^0.0.17`, `@orkestrel/ndjson` `^0.0.16`, `@orkestrel/emitter` `^0.0.15`, `@orkestrel/process` `^0.0.18`, `@orkestrel/contract` `^0.0.22` | |
70
+ | `@orkestrel/ollama` | `0.0.33` | L6 | `@orkestrel/tool` `^0.0.22`, `@orkestrel/agent` `^0.0.39`, `@orkestrel/budget` `^0.0.17`, `@orkestrel/ndjson` `^0.0.16`, `@orkestrel/emitter` `^0.0.15`, `@orkestrel/process` `^0.0.18`, `@orkestrel/contract` `^0.0.22` | |
71
71
  | `@orkestrel/pool` | `0.0.20` | L2 | `@orkestrel/emitter` `^0.0.15`, `@orkestrel/contract` `^0.0.22` | |
72
72
  | `@orkestrel/probe` | `0.0.29` | L5 | `@orkestrel/lsp` `^0.0.14`, `@orkestrel/mcp` `^0.0.42`, `@orkestrel/pool` `^0.0.20`, `@orkestrel/tool` `^0.0.22`, `@orkestrel/emitter` `^0.0.15`, `@orkestrel/process` `^0.0.18`, `@orkestrel/timeout` `^0.0.16`, `@orkestrel/contract` `^0.0.22` | `vitest` `^4.1.12 \|\| ^5.1.0`, `typescript` `^6.0.3 \|\| ^7.0.2` |
73
73
  | `@orkestrel/process` | `0.0.18` | L2 | `@orkestrel/emitter` `^0.0.15`, `@orkestrel/contract` `^0.0.22` | |
74
74
  | `@orkestrel/program` | `0.0.20` | L4 | `@orkestrel/rater` `^0.0.21`, `@orkestrel/reason` `^0.0.17`, `@orkestrel/emitter` `^0.0.15`, `@orkestrel/contract` `^0.0.22`, `@orkestrel/qualifier` `^0.0.21` | |
75
75
  | `@orkestrel/qualifier` | `0.0.21` | L3 | `@orkestrel/reason` `^0.0.17`, `@orkestrel/emitter` `^0.0.15`, `@orkestrel/contract` `^0.0.22` | |
76
- | `@orkestrel/queue` | `0.0.20` | L3 | `@orkestrel/abort` `^0.0.16`, `@orkestrel/emitter` `^0.0.15`, `@orkestrel/timeout` `^0.0.16`, `@orkestrel/contract` `^0.0.22`, `@orkestrel/database` `^0.0.21` | |
76
+ | `@orkestrel/queue` | `0.0.21` | L3 | `@orkestrel/abort` `^0.0.16`, `@orkestrel/emitter` `^0.0.15`, `@orkestrel/timeout` `^0.0.16`, `@orkestrel/contract` `^0.0.22`, `@orkestrel/database` `^0.0.22` | |
77
77
  | `@orkestrel/rater` | `0.0.21` | L3 | `@orkestrel/reason` `^0.0.17`, `@orkestrel/emitter` `^0.0.15`, `@orkestrel/contract` `^0.0.22` | |
78
78
  | `@orkestrel/reason` | `0.0.17` | L2 | `@orkestrel/emitter` `^0.0.15`, `@orkestrel/contract` `^0.0.22` | |
79
- | `@orkestrel/relation` | `0.0.19` | L3 | `@orkestrel/emitter` `^0.0.15`, `@orkestrel/contract` `^0.0.22`, `@orkestrel/database` `^0.0.21` | |
79
+ | `@orkestrel/relation` | `0.0.20` | L3 | `@orkestrel/emitter` `^0.0.15`, `@orkestrel/contract` `^0.0.22`, `@orkestrel/database` `^0.0.22` | |
80
80
  | `@orkestrel/router` | `0.0.20` | L2 | `@orkestrel/abort` `^0.0.16`, `@orkestrel/emitter` `^0.0.15`, `@orkestrel/contract` `^0.0.22` | |
81
- | `@orkestrel/scaffold` | `0.0.124` | L3 | `@orkestrel/console` `^0.0.19`, `@orkestrel/emitter` `^0.0.15`, `@orkestrel/process` `^0.0.18`, `@orkestrel/contract` `^0.0.22`, `@orkestrel/markdown` `^0.0.20`, `@orkestrel/template` `^0.0.13` | |
82
- | `@orkestrel/sea` | `0.0.21` | L3 | `@orkestrel/emitter` `^0.0.14`, `@orkestrel/process` `^0.0.17`, `@orkestrel/contract` `^0.0.21` | |
81
+ | `@orkestrel/scaffold` | `0.0.126` | L3 | `@orkestrel/console` `^0.0.19`, `@orkestrel/emitter` `^0.0.15`, `@orkestrel/process` `^0.0.18`, `@orkestrel/contract` `^0.0.22`, `@orkestrel/markdown` `^0.0.20`, `@orkestrel/template` `^0.0.13` | |
82
+ | `@orkestrel/sea` | `0.0.22` | L3 | `@orkestrel/emitter` `^0.0.15`, `@orkestrel/process` `^0.0.18`, `@orkestrel/contract` `^0.0.22` | |
83
83
  | `@orkestrel/server` | `0.0.26` | L3 | `@orkestrel/abort` `^0.0.16`, `@orkestrel/codec` `^0.0.9`, `@orkestrel/router` `^0.0.20`, `@orkestrel/emitter` `^0.0.15`, `@orkestrel/timeout` `^0.0.16`, `@orkestrel/contract` `^0.0.22` | |
84
84
  | `@orkestrel/sqlite` | `0.0.17` | L1 | `@orkestrel/contract` `^0.0.22` | |
85
85
  | `@orkestrel/sse` | `0.0.13` | L2 | `@orkestrel/codec` `^0.0.9`, `@orkestrel/contract` `^0.0.22` | |
86
- | `@orkestrel/study` | `0.0.11` | L7 | `@orkestrel/tool` `^0.0.22`, `@orkestrel/abort` `^0.0.16`, `@orkestrel/agent` `^0.0.36`, `@orkestrel/budget` `^0.0.17`, `@orkestrel/ollama` `^0.0.30`, `@orkestrel/timeout` `^0.0.16`, `@orkestrel/contract` `^0.0.22` | |
87
- | `@orkestrel/table` | `0.0.14` | L2 | `@orkestrel/emitter` `^0.0.14`, `@orkestrel/contract` `^0.0.21` | |
86
+ | `@orkestrel/study` | `0.0.16` | L7 | `@orkestrel/tool` `^0.0.22`, `@orkestrel/abort` `^0.0.16`, `@orkestrel/agent` `^0.0.39`, `@orkestrel/budget` `^0.0.17`, `@orkestrel/ollama` `^0.0.33`, `@orkestrel/timeout` `^0.0.16`, `@orkestrel/contract` `^0.0.22` | |
87
+ | `@orkestrel/table` | `0.0.15` | L2 | `@orkestrel/emitter` `^0.0.15`, `@orkestrel/contract` `^0.0.22` | |
88
88
  | `@orkestrel/template` | `0.0.13` | L2 | `@orkestrel/emitter` `^0.0.15`, `@orkestrel/contract` `^0.0.22` | |
89
- | `@orkestrel/terminal` | `0.0.23` | L3 | `@orkestrel/sse` `^0.0.13`, `@orkestrel/form` `^0.0.16`, `@orkestrel/console` `^0.0.19`, `@orkestrel/emitter` `^0.0.15`, `@orkestrel/contract` `^0.0.22`, `@orkestrel/database` `^0.0.21` | |
89
+ | `@orkestrel/terminal` | `0.0.24` | L3 | `@orkestrel/sse` `^0.0.13`, `@orkestrel/form` `^0.0.16`, `@orkestrel/console` `^0.0.19`, `@orkestrel/emitter` `^0.0.15`, `@orkestrel/contract` `^0.0.22`, `@orkestrel/database` `^0.0.22` | |
90
90
  | `@orkestrel/test` | `0.0.33` | L1 | `@orkestrel/contract` `^0.0.22` | `vitest` `^4.1.12 \|\| ^5.1.0`, `playwright` `^1.64.1` |
91
91
  | `@orkestrel/timeout` | `0.0.16` | L1 | `@orkestrel/contract` `^0.0.22` | |
92
92
  | `@orkestrel/tool` | `0.0.22` | L2 | `@orkestrel/emitter` `^0.0.15`, `@orkestrel/contract` `^0.0.22` | |
93
- | `@orkestrel/toolbox` | `0.0.25` | L6 | `@orkestrel/form` `^0.0.16`, `@orkestrel/tool` `^0.0.22`, `@orkestrel/agent` `^0.0.36`, `@orkestrel/server` `^0.0.26`, `@orkestrel/contract` `^0.0.22`, `@orkestrel/database` `^0.0.21`, `@orkestrel/relation` `^0.0.19`, `@orkestrel/terminal` `^0.0.23`, `@orkestrel/workflow` `^0.0.26`, `@orkestrel/workspace` `^0.0.15` | |
93
+ | `@orkestrel/toolbox` | `0.0.28` | L6 | `@orkestrel/form` `^0.0.16`, `@orkestrel/tool` `^0.0.22`, `@orkestrel/agent` `^0.0.39`, `@orkestrel/server` `^0.0.26`, `@orkestrel/contract` `^0.0.22`, `@orkestrel/database` `^0.0.22`, `@orkestrel/relation` `^0.0.20`, `@orkestrel/terminal` `^0.0.24`, `@orkestrel/workflow` `^0.0.27`, `@orkestrel/workspace` `^0.0.16` | |
94
94
  | `@orkestrel/websocket` | `0.0.18` | L2 | `@orkestrel/emitter` `^0.0.15`, `@orkestrel/contract` `^0.0.22` | |
95
- | `@orkestrel/worker` | `0.0.19` | L4 | `@orkestrel/pool` `^0.0.19`, `@orkestrel/queue` `^0.0.19`, `@orkestrel/emitter` `^0.0.14`, `@orkestrel/contract` `^0.0.21`, `@orkestrel/database` `^0.0.20` | |
96
- | `@orkestrel/workflow` | `0.0.26` | L4 | `@orkestrel/abort` `^0.0.16`, `@orkestrel/queue` `^0.0.20`, `@orkestrel/budget` `^0.0.17`, `@orkestrel/emitter` `^0.0.15`, `@orkestrel/timeout` `^0.0.16`, `@orkestrel/contract` `^0.0.22`, `@orkestrel/database` `^0.0.21` | |
97
- | `@orkestrel/workspace` | `0.0.15` | L3 | `@orkestrel/emitter` `^0.0.15`, `@orkestrel/contract` `^0.0.22`, `@orkestrel/database` `^0.0.21` | |
95
+ | `@orkestrel/worker` | `0.0.21` | L4 | `@orkestrel/pool` `^0.0.20`, `@orkestrel/queue` `^0.0.21`, `@orkestrel/emitter` `^0.0.15`, `@orkestrel/contract` `^0.0.22`, `@orkestrel/database` `^0.0.22` | |
96
+ | `@orkestrel/workflow` | `0.0.27` | L4 | `@orkestrel/abort` `^0.0.16`, `@orkestrel/queue` `^0.0.21`, `@orkestrel/budget` `^0.0.17`, `@orkestrel/emitter` `^0.0.15`, `@orkestrel/timeout` `^0.0.16`, `@orkestrel/contract` `^0.0.22`, `@orkestrel/database` `^0.0.22` | |
97
+ | `@orkestrel/workspace` | `0.0.16` | L3 | `@orkestrel/emitter` `^0.0.15`, `@orkestrel/contract` `^0.0.22`, `@orkestrel/database` `^0.0.22` | |
98
98
 
99
99
  <!-- /orkestrel:catalog -->
100
100
 
@@ -1,13 +1,13 @@
1
1
  ---
2
2
  name: planner
3
- description: 'Opus 5.5 design and consultation lane by default, read-only: proposes a coherent API shape, vocabulary, architecture, alternatives, and bounded units for one design brief, or gives a consultation at a seam, including the Fable 5.1 consultation on the engine the dispatch names. Holds the subjective lane by default and the objective lane when the dispatch assigns it. Never implements or accepts.'
3
+ description: 'Opus 5.5 design and consultation lane by default, read-only: proposes a coherent API shape, vocabulary, architecture, alternatives, and bounded units for one design brief, or gives a consultation at a seam, including the Fable 5.2 consultation on the engine the dispatch names. Holds the subjective lane by default and the objective lane when the dispatch assigns it. Never implements or accepts.'
4
4
  tools: Read, Grep, Glob
5
5
  model: opus
6
6
  effort: high
7
7
  permissionMode: plan
8
8
  ---
9
9
 
10
- On Claude Opus 5.5 by default, or on the engine the dispatch names, design a dispatched unit or give a read-only consultation under `.agents/orchestration.md` § Sessions and consultation. You also carry the read-only consultation `.agents/orchestration.md` § Engines admits for Claude Fable 5.1, on the engine the dispatch names. Run no command and edit nothing; return the proposal or consultation in your final message.
10
+ On Claude Opus 5.5 by default, or on the engine the dispatch names, design a dispatched unit or give a read-only consultation under `.agents/orchestration.md` § Sessions and consultation. You also carry the read-only consultation `.agents/orchestration.md` § Engines admits for Claude Fable 5.2, on the engine the dispatch names. Run no command and edit nothing; return the proposal or consultation in your final message.
11
11
 
12
12
  ## Do
13
13
 
@@ -29,6 +29,7 @@ These rules govern how a claim is established: the evidence read before a change
29
29
  - A clean reproduction of a reported defect means your input was weaker than the reporter's; get theirs before ruling. An input the compiler rejects refutes that input, not the finding.
30
30
  - "No tests found", an empty match, or a runner that resolved nothing reports on the harness. Confirm the probe was collected before reading a result.
31
31
  - Prefer an observation to a derivation. When a measurement and an argument disagree, the argument is wrong until the measurement is shown broken.
32
+ - When a probe compares a candidate with its base, extract the base with the `git archive` command into its own tree under `tmp/probes/`, link the checkout's `node_modules` directory into that tree, run the same probe in both trees, and read each arm from the JSON file it writes.
32
33
  - Read a gate bare. A `| tail` or `| grep` behind it hides the failing lines and reports the pipe's exit code.
33
34
 
34
35
  ## Instruments
@@ -45,6 +46,7 @@ These rules govern how a claim is established: the evidence read before a change
45
46
  - Read a receipt as evidence about its claim, never as a gate result.
46
47
  - State the instrument's coverage beside its result. Match the instrument to the question: a text search reports on text; a claim about declarations or call sites needs the compiler or a parser.
47
48
  - Report a question unanswered rather than answering it with a weaker instrument.
49
+ - When a verdict reads rows or numbers, refuse the verdict on an empty result or on a missing or non-finite number; never read either as zero or as a pass.
48
50
  - Baseline a published-artifact claim against the published artifact (the tarball, the deployed asset). Prove a module cycle by loading the built entry points.
49
51
  - Promote an instrument that settled a claim into a test, with its control, before accepting the work it settled.
50
52
 
@@ -55,6 +57,7 @@ These rules govern how a claim is established: the evidence read before a change
55
57
  - Refuse a micro optimization (hoisting a small array, caching one cheap read, shaving a constant) unless its own measurement shows a gain worth having.
56
58
  - Keep an algorithmic repair (a superlinear term, a repeated layout) when the measurement shows it.
57
59
  - Before the run, size the fixtures, the sample count, and the threshold to the case, and state the reasoning and the result that makes the change worth keeping.
60
+ - Before setting a floor on a quality or cost reading, read the spread its setting shows under a small unrelated change, and set the floor outside that spread.
58
61
  - Read the difference between methods against each side's reported uncertainty, and record no difference smaller than the result stated before the run.
59
62
  - Report every measurement with the host's other load during the run, its memory and console counts included.
60
63
  - When a brief changes a hot path, name a measured cost bar at the input limit, and read the cost there before acceptance: an ordinary-length test cannot read a superlinear term.
@@ -37,11 +37,13 @@ These rules govern what a test asserts and where it lives, the cross-cutting and
37
37
  - Use recorders for calls/events, temporary resources for stateful boundaries, protocol-faithful fixture servers for deterministic network peers, and the real external service when its behavior is the claim.
38
38
  - Prefer inert customizable data and input stubs.
39
39
  - Allow a scripted boundary stub only when it implements the real interface or protocol minimally, to drive the system under test. It never reimplements project-owned behavior and never stands in for the integration being claimed.
40
+ - When a dependency's release changes a wire a scripted boundary stub serves, update the stub to the wire as the dependency sends it, and keep every caller-facing assertion.
40
41
  - Bind a test fixture server to `127.0.0.1` on an ephemeral port (`listen(0)`), never to `::1` and never to a fixed port: a host without IPv6 fails `EAFNOSUPPORT`, and a fixed port flakes on occupancy.
41
42
  - Cover happy paths, error paths, empty input, boundary values, `NaN`, positive/negative zero, cycles, and Map/Set order where relevant.
42
43
  - Test observable behavior, not implementation details.
43
44
  - Assert the membership a discovered or globbed set must have, not a total that a partly empty population satisfies. A glob spanning two locations passes a size check while one of them matches nothing.
44
45
  - Never assert an implementation against itself. Compare the answer to a declaration, a fixture, or a second mechanism that could disagree with it. Re-deriving the answer the same way the source derives it produces a test that passes for every value the source ever returns, and it reads exactly like a real one.
46
+ - When two mechanisms agree, also drive the real harness's call to them on an input whose answer is known: their parity reads neither one's call site.
45
47
  - Probe a host-varying property at runtime, on the host the test is running on, and assert against what the probe returned. Filesystem case folding, path separators, permission bits, and rename semantics differ per host, and a thrown error's class or message differs per Node major, so a fixture built on one host describes that host and silently measures something else on the next.
46
48
  - Assert a runtime-chosen result as the property it must have, not as the number one run produced. Compression, timing, and buffer sizing are the runtime's choice, so pin the relationship the test depends on — that the encoded form is larger, that the second call is faster — and let the assertion fail when the input drifts out of the range where that relationship holds.
47
49
  - Measure an elapsed interval with `performance.now()`, never `Date.now()`. `Date.now()` returns whole milliseconds, so an interval built from two of its readings truncates at both ends and can under-report by a millisecond — enough to fail a boundary assertion against a timer that behaved correctly. `performance.now()` is monotonic and sub-millisecond, and it does not move when the wall clock does.
@@ -347,6 +349,7 @@ Before accepting that a behavior cannot be tested, look for the seam that would
347
349
 
348
350
  - Treat a collaborator reached through a hard-coded global — a stream, a clock source, a spawn, a fetch — as a missing injection point, not an untestable truth.
349
351
  - An injected collaborator with a real minimal implementation is a sanctioned boundary stub, not a mock of project-owned behavior.
352
+ - When a recording replays a wire, keep run-local identity, such as a random id or a clock reading, off it: after reading on the real peer that it treats the field as a label, send a key every run sends alike, such as the item's position, and map the answers back at the caller.
350
353
  - Add the seam rather than shipping the gap whenever the design would welcome it anyway, and whenever a sibling collaborator is already injected.
351
354
  - Record a genuinely irreducible gap where a reader meets it: what is unproven, why it cannot be driven, and what would change that. A silent untested guard reads exactly like a tested one.
352
355
 
@@ -166,7 +166,8 @@ environment axis is one project per src/app axis × environment, plus one per ex
166
166
  the one root `sheetProject` factory, and run it through its `test:src:<face>` script, which builds
167
167
  that face first.
168
168
  - Give every browser project `optimizeDeps.include` of `@orkestrel/test`, `@orkestrel/test/browser`,
169
- `@orkestrel/contract` where the manifest declares it, and `vue` where the project renders Vue.
169
+ `@orkestrel/contract` where the manifest declares it, the `/browser` entry of every declared
170
+ `@orkestrel` package that publishes one, and `vue` where the project renders Vue.
170
171
  Give a Node project no `optimizeDeps` setting.
171
172
  - Give every browser project `test.browser.commands` from `configs/commands.ts`: each root factory
172
173
  that names the Playwright provider reads the record, and a project that spreads another
@@ -185,7 +186,7 @@ one environment, so each is its own project:
185
186
  | `config` | `tests/config.test.ts` | Root configuration resolves its aliases, projects, and outputs, and the `configs/` leaves behind them | `test` |
186
187
  | `setup` | `tests/setup*.test.ts` other than either `setup:browser` proof | Prove root setup behavior in Node with `setup.ts`. | `test` |
187
188
  | `setup:browser` | `tests/setupBrowser.test.ts`, `tests/setupStyles.test.ts` | Prove browser and style setup behavior in Playwright Chromium with `setup.ts` and `setupBrowser.ts`. | `test` |
188
- | `journey:<variant>` | `tests/app/<application>/integration.test.ts` | Drive the application the Vite mode selects at the declared variant viewport in Playwright Chromium with `setup.ts` and `setupBrowser.ts`. | `test` through the journey scripts |
189
+ | `journey:<variant>` | `tests/app/<application>/integration.test.ts` | Drive the application the Vite mode selects at the declared variant viewport in Playwright Chromium with `setup.ts` and `setupBrowser.ts`; every browser factory that loads `setupBrowser.ts`, `srcBrowser`, `appBrowser`, and `sheetProject` included, carries `vue()` when the workspace has a Vue face. | On demand; `test` when private |
189
190
  | `guides` | `tests/guides.test.ts` | Every documented API exists, every public API is documented, every compared summary, example, and pitch equals its source, and every executable fence returns what the guide says it returns | `test` |
190
191
  | `conformance` | `tests/conformance.test.ts` | Where this package drifts from the official tooling it tracks | `test` |
191
192
  | `tree` | `tests/tree.test.ts` | The laws this package states about its own real tree that no vendored proof carries | `test` |
@@ -215,8 +216,8 @@ one environment, so each is its own project:
215
216
  factory, which resolves the Vite mode to `browser` or an app-side browser extension and collects
216
217
  `tests/app/<application>/integration.test.ts` for that application. The factory merges its
217
218
  `override` last, through `mergeOverride`, so an override replaces nothing it does not name. Exclude each collected suite
218
- from its application project, and run `test:journey` and one `test:journey:<framework>` per
219
- app-side browser extension after the application projects in `test`.
219
+ from its application project. Generate `test:journey` when the workspace has the base entry `app/browser/main.ts`, and one `test:journey:<framework>` per
220
+ app-side browser extension, and place them by the gate-placement paragraph that follows.
220
221
 
221
222
  Keep `conformance`, `tree`, `integration`, `distribution`, `studies`, and `service` as separate
222
223
  subjects.
@@ -230,6 +231,11 @@ parallelism in `service`. In a `private: true` workspace, never declare `prepubl
230
231
  `distribution`, reach `service` from `test`, and retain each isolated project's configuration.
231
232
  In both kinds of workspace, run `studies` on demand through `npm run test:studies` alone.
232
233
  Keep `test:studies` out of every gate and chain.
234
+ In a publishing workspace, run application and journey suites on demand through `test:app`,
235
+ `test:app:<environment>`, `test:app:<framework>`, `test:journey` (with the base entry), and
236
+ `test:journey:<framework>`. Keep them out of every release gate and chain.
237
+ In a `private: true` workspace, keep `test:app` in `test`, followed by `test:journey` (with the base entry) and
238
+ each `test:journey:<framework>` the workspace declares.
233
239
 
234
240
  One project sits on neither axis. `probe` includes `tmp/probes/**/*.test.ts` so an agent can run a
235
241
  throwaway instrument against real sources, aliases and setup. Declare no proof there. Keep the
@@ -322,7 +328,7 @@ Build/check config alignment:
322
328
  | `prepublishOnly` | Publishing workspaces only: the gate chain, then the isolated release proofs |
323
329
  | `prepack` | Publishing workspaces only: rebuild `dist/` so a pack ships current output |
324
330
 
325
- - In a publishing workspace, `prepublishOnly` runs `build:showcase` and every
331
+ - In a publishing workspace, `prepublishOnly` runs `build:showcase` when the workspace has the base entry, and every
326
332
  `build:showcase:<framework>` script after `npm run build`.
327
333
  - List `showcase/` and `*.min.*` in `.prettierignore`, so formatting never rewrites a committed page or a minified generated record. Give the `.min` suffix only to a record an instrument writes from an installed package in compact form (a JSON inventory over 1 MB, for example), never to authored source, and read such a record through a guarded reader. Never list a source file there; `.claude/rules/styles.md` § Prohibitions states how a pinned recreation stays formatter-stable.
328
334
 
@@ -1,12 +1,12 @@
1
1
  name = "planner"
2
- description = "Read-only GPT-6 Astra design of shape, naming, ergonomics, alternatives, and bounded units, or a consultation at a seam, including the Fable 5.1 consultation on the engine the dispatch names. Never implements or accepts."
2
+ description = "Read-only GPT-6 Astra design of shape, naming, ergonomics, alternatives, and bounded units, or a consultation at a seam, including the Fable 5.2 consultation on the engine the dispatch names. Never implements or accepts."
3
3
  model = "gpt-6-astra"
4
4
  model_reasoning_effort = "high"
5
5
  sandbox_mode = "read-only"
6
6
  developer_instructions = """
7
7
  Design one dispatched unit or give a read-only consultation under .agents/orchestration.md
8
8
  § Sessions and consultation on GPT-6 Astra. Also carry the read-only consultation
9
- .agents/orchestration.md § Engines admits for Claude Fable 5.1, on the engine the dispatch names.
9
+ .agents/orchestration.md § Engines admits for Claude Fable 5.2, on the engine the dispatch names.
10
10
  Use read-only inspection commands. Never edit, run a mutating command, or spawn.
11
11
 
12
12
  Read the brief, AGENTS.md, the rules whose paths match the subject, the named skill, the guide,