Skip to content

feat: one governance seam and neutral naming for decision models - #36267

Closed
bolichen97 wants to merge 1 commit into
mainfrom
feat/decision-model-governance
Closed

bolichen97 wants to merge 1 commit into
mainfrom
feat/decision-model-governance

Conversation

@bolichen97

Copy link
Copy Markdown
Contributor

Problem / Motivation

Goal: An administrator can turn decision models off entirely or allow only specific ones, enforced by the gateway, and the dashboard calls the feature "decision model" instead of naming one provider.

Today the feature is named after one provider ("Decisions (Jev)", "Jev API key", "Share of sessions Jev answers for", "Jev chose: ...") even when a local model or a SageMaker endpoint answers, and governance knows only two special cases: the hosted host and the SageMaker host. An administrator cannot allow Laya and refuse everything else, and a policy that turns the row off makes the card disappear with no word of why. The hosted provider's API key sits in a "Shared settings" block although only that provider uses it.

Why it matters

A managed fleet (the internal edition first) must be able to ship with one provider not allowed, and every user must be able to read why a model is unavailable. Naming the feature after one provider reads wrong for every user whose decisions are answered by another model.

Not a goal

  • Editing the internal edition's companion package. This PR provides the seam; the exact setting the edition ships is listed under "What changed".
  • Renaming any stored identifier, config key, API field or ceiling key: jev stays the preset id and lane value, auto:jev the slot model, catalog key names (auto_jev, rate_jev) unchanged.
  • Putting the wake judge's small-model lane in the allow list: it runs on the session's own model provider and stays bound by the row and the points list, as models and endpoints already are.
  • Naming the provider that answered on a transcript decision row: the record carries no provider id, and the currently active model would mislabel every row decided before a switch, so the rows say "Decision model".
  • Changing CONTRACT_VERSION, the evaluator, or the standalone default (no policy: every model allowed).

What changed (motivation → approach → change)

Approach. Generalize the existing capabilities.decisions ceiling instead of adding a parallel mechanism: one more inner list on the same row, asked wherever the row is already asked for the configured provider.

Policy keys and data shape.

  • capabilities.decisions.enabled stays the full off switch.
  • New inner list capabilities.decisions.scopes.providers (identifier matcher, a SCOPE_CATALOG data row). One entry per decision model, by the picker's own id: jev (hosted), plumb-4b, strands-decider-2b, laya, sagemaker (any sagemaker:// address), custom (any other hand-written address). "No model" is never asked. No list means every model is allowed.

Enforcement. capability.provider_id(endpoint, model) names the configured model; capability.is_decisions_denied(provider=...) asks providers:<id> together with the row(s) inside one policy-generation bracket (fails closed if the policy moved between the reads). It is asked at every chokepoint that reads the row for the configured provider:

  • send path: gate._consented_for, so decide, is_enabled and warm refuse a refused model whichever writer named it in config.json;
  • PUT /api/decisions/provider (switch_refused): 403 decisions_capability_denied, nothing written;
  • PUT /api/decisions/consent: an enabling or scope-granting write for a refused model is 403; a disabling write is never refused;
  • the consent carry to a local preset, the SageMaker connection test, and local_runtime.resume_configured (a refused preset is not started).

The keystone is never touched by a refusal: the recorded consent stays in place and inert, and counts again if a later policy allows that model.

Settings read. GET /api/decisions/provider adds provider_permitted: {id: bool}, each the AND of the model's row, the endpoints list for a model that leaves the machine, and its own entry (dashboard/handlers/decisions_offers.py, asked without a SEL row per id since it is an offer; every request that acts is audited). hosted_permitted / sagemaker_permitted / configured_refused fold the list in, so a configured hosted default the list refuses reads as "No model" with no config rewrite.

UI.

  • Off switch: the card draws its name and one plain notice, "Your administrator has turned off decision models on this machine. ...", with no switch and no picker (role="status"). Previously the card vanished.
  • Allow list: every refused model is drawn greyed, unselectable, with "Not allowed by your administrator."; the hosted provider and SageMaker are no longer hidden. When every option is refused the picker says so; "No model" is always choosable. A refused hand-written address is named under the card's address pointer.
  • Naming: "Decision model" in every label, hint, switch, title and error, in all 12 catalogs. The hosted provider's entry ("Jev, hosted by TypeSafe") is the only place Jev is named; it now holds its own TypeSafe API key field (DecisionsHostedKeyField.tsx), opened when the entry is chosen or a settings search targets the key. The shared block keeps only the shared settings (share of sessions, history budget). Old settings-search ids are rewritten to the new ones (useSettingHighlight legacy table).
  • Backend strings a user reads were neutralized too (the consent and provider refusals, the owner-only model refusals, the MCP session_set_model description, three config-schema descriptions); Auto (decision model) is refused for a non-owner like the old label spellings.

Specs and docs, same commit. docs/system-specs/modules/decisions.md (naming rule, section 7 and new 7.2), governance.md (the providers list), the feature map and spec label quotes, and the user docs under src/kiro_crew/docs/ (new "When your administrator limits decision models" section). No entry in docs/decisions/ covers these labels or this placement, so no superseding entry was needed.

What the internal edition sets (requirement 4). In the companion's bundled security_policy.json (the edition's posture file, precedence tier "bundled"), add under the existing top level:

  • file: the internal companion's bundled security_policy.json
  • key: capabilities.decisions.scopes.providers
  • value: {"mode": "deny", "deny": ["jev"]}
{"version": 1, "capabilities": {"decisions": {"scopes": {"providers": {"mode": "deny", "deny": ["jev"]}}}}}

If the bundled policy already names capabilities.decisions, merge the scopes.providers key into that object (do not add "enabled" unless the edition means to turn the feature off; for the full off switch it is capabilities.decisions.enabled = false). A governance profile bound to a surface can carry the same scopes.providers object to narrow that surface only. This exact fragment is pinned by test_decisions_provider_allowlist.py::TestTheEditionSnippet.

Inventory of the naming sweep (greps for Jev, jev, JEV, System One, TypeSafe, typesafe, PRESET_JEV across website/src, every catalog, the generated UI and settings indexes, backend strings that reach the UI, CLI and MCP text, and src/kiro_crew/docs/):

  • (a) feature-level wording, made neutral: 72 catalog keys rewritten in all 12 catalogs plus 3 new keys; 9 backend strings (2 handler refusals, 1 SageMaker-test refusal, 2 owner-only model refusals, 1 MCP tool description, 3 config-schema descriptions); 5 settings-search registry labels and ids; 4 user docs files; label quotes in the feature map and 3 specs.
  • (b) kept on purpose, all inside or about the hosted provider's own entry: decisions_provider_jev ("Jev, hosted by TypeSafe", the only catalog value naming Jev); decisions_provider_jev_detail (names TypeSafe); decisions_api_key_label ("TypeSafe API key", drawn inside the entry); decisions_hosted_refused (interpolates the provider's name and is drawn inside the entry); decisions_custom_loopback_no_key (names the TypeSafe key that is not sent to a loopback address); "System One" in the SageMaker entry's detail and its connection-test protocol error (the wire format that endpoint must speak).
  • (c) not user-facing, unchanged so saved settings and logs keep working: about 566 lines in 39 frontend source files and about 301 lines in 55 backend modules (the jev preset id and lane value, auto:jev, PRESET_JEV, jevRoute.ts, useJevAutoSend, catalog key names, TYPESAFE_API_KEY, jev-latest, log lines, comments).

Backwards compatibility

Compatible: no policy, or a policy without the providers list, allows every model, so every saved provider selection, consent and config keeps working; no config key, API field or ceiling key is renamed or removed (provider_permitted is added; old settings-search ids are rewritten to the new ones).

The diff tightens what a policy can refuse and changes what the card draws under a ceiling, so its readers:

Reader: src/kiro_crew/decisions/gate.py:_consented_for -- chat -- test_decisions_provider_allowlist.py::TestTheGateHonoursTheList::test_no_policy_sends_to_every_saved_selection
Reader: src/kiro_crew/decisions/gate.py:_consented_for -- cron -- the autonudge judge reads the same consent; test_decisions_judge_llm.py passes unchanged
Reader: src/kiro_crew/decisions/gate.py:_consented_for -- subagent -- browser.step and spawned-task routing funnel through the same read; test_browser_reflex.py passes unchanged
Reader: src/kiro_crew/dashboard/handlers/decisions.py:_carry_consent -- chat -- test_decisions_local_provider.py::TestProviderRoutePut::test_two_switches_never_interleave_write_and_carry
Reader: src/kiro_crew/dashboard/handlers/decisions_sagemaker.py:switch_refused -- chat -- test_decisions_sagemaker_provider.py::test_a_fleet_denial_of_the_hosted_side_refuses_it
Reader: src/kiro_crew/dashboard/handlers/decisions.py:_provider_payload -- chat -- with no list the side answers are unchanged; test_decisions_sagemaker_provider.py::TestRefusedOptionIsNotOffered
Reader: src/kiro_crew/decisions/local_runtime.py:resume_configured -- release -- runs at gateway start; test_decisions_provider_allowlist.py::TestTheRuntimeResume, test_decisions_local_runtime.py
Reader: src/kiro_crew/dashboard/session_control_parts/model_switch.py:_JEV_ROUTE_SPELLINGS -- subagent -- old spellings kept, the new label added; test_session_control_set_model.py::test_auto_is_owner_only
Reader: website/src/pages/settings/DecisionsCard.tsx:DecisionsCard -- chat -- a denied row draws the notice instead of nothing; decisionsCard.test.tsx "says the administrator turned decision models off, and offers no control"
Reader: website/src/pages/settings/DecisionsProviderPicker.tsx:DecisionsProviderPicker -- chat -- refused options shown greyed instead of hidden; decisionsProviderPicker.test.tsx
Reader: website/src/hooks/useSettingHighlight.ts:resolveLegacyHighlightId -- chat -- old ids (developer.decisions-jev, developer.jev-api-key, three scope switches) resolve to the new ones; settingHighlightAliases.test.ts
Reader: src/kiro_crew/guide_catalog.py:_REGISTRY_PATH -- chat -- regenerated registry; test_guide_ui_find.py, test_find_ui_heldout.py

Tests

  • test/test_decisions_provider_allowlist.py (new, real policy and evaluator): the catalog row; provider_id per address shape; at the gate, no policy and a list-less policy send, a refused model (deny, allow-other, empty allow) sends nothing and leaves the keystone byte-identical, consent counts again once allowed, a hand-written custom address answers to its own entry, a local preset to its own entry, the off switch refuses all, a policy swap between row and entry denies, "none" is never asked; the provider route (per-model GET, 403 and no write for refused jev, preset, SageMaker, "none" never refused); the consent route (403 for a refused model, disabling allowed, GET reports permits: false with consent shown as recorded); runtime resume; the edition snippet.
  • Mutation-tested: dropping the provider from gate._consented_for, switch_refused, _seam_withdrawn, resume_configured, forcing permitted_providers true, removing the generation bracket, and mis-classifying SageMaker each fail the file.
  • website/src/i18n/decisionModelNaming.test.ts (new): every catalog names Jev only under the hosted provider's own keys, with a self-check sample. website/src/pages/settings/decisionModelNaming.test.tsx (new): the card in each state (all allowed with the hosted entry and key open, an allow list of one, every model refused, SageMaker active, a local model active, turned off by the administrator) and the transcript rows (steer line, skill strip) under each active provider: no text, aria-label or title outside the hosted entry names Jev. Mutation-tested: reintroducing "Jev" into a catalog value, dropping the entry marker, dropping the notice, or ignoring provider_permitted each fail.
  • decisionsProviderPicker.test.tsx: the key field lives inside the hosted entry only, opens on a search hit, and is not drawn for a refused provider; refused options are greyed, not hidden. Existing card, scope, strip and picker tests updated to the new labels.
  • Lint gates run locally: pinned black gate, isort 6.0.0, flake8 7.1.0, pinned mypy 1.14.1 --platform linux (clean), comment-history, brand-name, file line cap, harness parity, flake-pattern ratchet, npm run i18n:check with the merge base (all rows pass), tsc -p tsconfig.app.json, eslint on touched files, jscpd.

Manual verification

Ran the real built SPA through the screenshot harness (website/scripts/capture-decision-model-governance.mjs, shared serve-dist and stub-dashboard-api helpers) against this branch and against a build of origin/main (--before), with every /api/** call answered from fixtures; the after run asserts each state's text and that nothing outside the hosted entry names Jev. The steer trace line was captured with the existing capture-decision-message-steer.mjs on both builds. No gateway or live provider was used.

Screenshots / video

Committed under temp-screenshots/decision-model-governance/ (light theme):

State Before After
All models allowed, hosted entry chosen temp-screenshots/decision-model-governance/before/all-allowed.png temp-screenshots/decision-model-governance/after/all-allowed.png
Allow list with one model (Laya) temp-screenshots/decision-model-governance/before/allow-list-one.png temp-screenshots/decision-model-governance/after/allow-list-one.png
Turned off by the administrator temp-screenshots/decision-model-governance/before/admin-disabled.png (no card) temp-screenshots/decision-model-governance/after/admin-disabled.png
SageMaker endpoint active temp-screenshots/decision-model-governance/before/sagemaker-active.png temp-screenshots/decision-model-governance/after/sagemaker-active.png
Local model active temp-screenshots/decision-model-governance/before/local-active.png temp-screenshots/decision-model-governance/after/local-active.png
Transcript trace row (queued) temp-screenshots/decision-model-governance/before/trace/01-queued-light.png temp-screenshots/decision-model-governance/after/trace/01-queued-light.png
Transcript trace row (steered) temp-screenshots/decision-model-governance/before/trace/02-steered-light.png temp-screenshots/decision-model-governance/after/trace/02-steered-light.png
Send-button picker temp-screenshots/decision-model-governance/before/trace/03-picker-light.png temp-screenshots/decision-model-governance/after/trace/03-picker-light.png

After, turned off by the administrator:

after: turned off by the administrator

After, transcript trace row (before it read "Jev chose: run after this turn"):

after: trace row

After: allow list of one model, and all models allowed with the hosted entry and its key open

after: allow list of one

after: all allowed

Related Issues

Maintainer request (Bolin Chen, bolichen97), 2026-11-12: one governance seam over all decision models; the internal edition ships with the hosted provider not allowed.

Checklist

  • At most two commits (one is the norm), with a Conventional Commits title (feat|fix|docs|style|refactor|perf|test|chore|ci|build|revert: ...)
  • Existing tests pass and new tests added for new functionality
  • Self-review completed; code follows project style guidelines
  • Documentation updated (if applicable)
  • No secrets, credentials, or internal references in the diff

Contribution License Agreement

Contributed under the project's Apache-2.0 license.

Decisions were named after one provider, and governance knew only two
special cases (the hosted host and SageMaker). Every decision model now
has its own entry in a per-model allow list, and the UI names the
feature, not the provider.

- capabilities.decisions gains a `providers` inner list (SCOPE_CATALOG
  data, no evaluator change). Ids are the picker's own: jev, each local
  preset, sagemaker, custom. The configured model's entry is asked with
  the row in one policy-generation bracket at every check that reads the
  row for it: the send path, the consent and provider writes, the
  consent carry, the SageMaker test and the runtime resume. A refused
  model is neither selectable (403) nor used, whoever wrote config.json;
  its recorded consent stays in place and inert. No policy, or a policy
  without the list, allows every model, so saved selections keep working.
- GET /api/decisions/provider reports provider_permitted per model. The
  picker draws a refused model greyed with "Not allowed by your
  administrator." instead of hiding it, and the card draws a plain
  "turned off by your administrator" notice when the row is off.
- "Decision model" is the feature in every label, hint, error and
  catalog; Jev is named only inside the hosted provider's own entry,
  which now holds its TypeSafe API key field. Stored ids (jev, auto:jev,
  catalog key names) are unchanged. Catalog and render tests pin it.
- Specs (decisions, governance) and the user docs describe the list,
  the notice and the new labels; before and after screenshots are under
  temp-screenshots/decision-model-governance.
@bolichen97
bolichen97 requested review from a team and KiroCrewCaptain as code owners November 13, 2026 01:17
@github-actions github-actions Bot added the readiness: checking Automated validation is still running label Nov 13, 2026
@github-actions

Copy link
Copy Markdown
Contributor

First Principles Review (Opus 5.5) — ✅ PASS

Premise-level review of 23655f1d17480fcba37e572278fb4d14ec50f0bd — why this exists and whether the shipped surface is the smallest honest version. Updated in place on each push. A BLOCK verdict blocks PR readiness; PASS does not.

First-Principles-Verdict: PASS

Nothing blocks. One note: the hosted provider's API key field moves into that provider's picker entry, and the PR doesn't name anyone who couldn't find it where it was.

Notes

  • Item 7 — unjustified move: the only harm stated is "The hosted provider's API key sits in a 'Shared settings' block although only that provider uses it." That says where the key belongs, but not who failed to find it. Existing users now have to relearn where the key lives, and the old search ids need a redirect table to keep working. Subtraction: delete DecisionsHostedKeyField.tsx and keep the key field in the shared block, renamed "TypeSafe API key"; then drop the developer.jev-api-key redirect row in useSettingHighlight.ts.

What this change ships

Inventory (10 items) — 9 justified

Intent: an administrator can switch decision models off or allow only some of them, enforced by the gateway, and every user sees the feature by a neutral name and can read why a model is unavailable. ADDITION.

  1. Administrators can allow or refuse each decision model through capabilities.decisions.scopes.providers — justified
  2. A refused model is blocked when sending, when switching provider, on consent writes, on local resume and on the SageMaker test — justified
  3. The settings read reports whether each model is allowed (provider_permitted) — justified
  4. When the off switch is set, the card stays visible with an administrator notice instead of disappearing — justified
  5. Refused hosted and SageMaker options are drawn greyed with "Not allowed by your administrator" instead of being hidden — justified
  6. "Jev" becomes "Decision model" in labels, errors, the MCP tool description and the docs — justified
  7. The hosted API key moves from Shared settings into the hosted entry, renamed "TypeSafe API key" — unjustified move
  8. Old settings-search deep links redirect to the new ids — justified
  9. A non-owner choosing "Auto (decision model)" is refused, like the old label spellings — justified
  10. Before/after screenshots added and capture scripts updated — justified

[FIRST-PRINCIPLES-REVIEWED] 23655f1

Description read

[DESCRIPTION-READ] e3174aeffae1424c0757faab6fd8cf73b65a3d16377d2e598c2264dfc4cdcdf0

That sha256 names the PR description this verdict read. Recompute it with .github/scripts/pr-description-capture.sh; if it differs, the description changed after this verdict was formed and any finding drawn from the description is unproven.

@bolichen97
bolichen97 enabled auto-merge (squash) November 13, 2026 01:20
@github-actions

Copy link
Copy Markdown
Contributor

Opus 5.5 Review — 🔴 changes requested (blocking)

Reviewed 23655f1d17480fcba37e572278fb4d14ec50f0bd — this comment is updated in place on each push.

Blocked: a settings-search hit on the hosted API key opens the key field and then closes it again, so search can't be used to set the key.

BLOCKING — website/src/pages/settings/DecisionsProviderPicker.tsx:443
selected ?? (openHostedKey && !blockedId(PRESET_JEV) ? PRESET_JEV : configuredChoice)
Trigger: the configured decision model is a local preset, SageMaker or "No model", and the user picks "TypeSafe API key" in settings search. That gives ?highlight=developer.typesafe-api-key, so DecisionsCard computes hostedKeyTargeted (DecisionsCard.tsx:198) from the URL on every render and passes it as openHostedKey, which makes chosen the hosted entry and mounts DecisionsHostedKeyField. The registry entry has no settingId or configKey, so useSettingHighlight takes the label path: after 100ms it flashes the label and always strips highlight.
Outcome: on the next render openHostedKey is false and selected is still null, so chosen goes back to configuredChoice and the key field unmounts. The search hit flashes the field and then removes it, so the user can't enter the key from search.
Anchor: defect
Fix: latch the search target the first time openHostedKey is true. For example, call setSelected(PRESET_JEV) in an effect, or keep a useState seeded from the prop, so chosen no longer depends on the URL param after it is stripped.

[BLOCK-MERGE] 23655f1
[OPUS-REVIEWED] 23655f1

Verdict parsed from the review's SHA-scoped output markers for commit 23655f1d17480fcba37e572278fb4d14ec50f0bd.

False positive or not applicable? A repository writer can comment:
/ai-review override fable 23655f1d17480fcba37e572278fb4d14ec50f0bd false-positive: <reason, 20 to 500 characters>

@github-actions

Copy link
Copy Markdown
Contributor

GPT 6.1 Review — 🔴 changes requested (blocking)

GPT 6.1 found at least one blocking issue that must be resolved before merging 23655f1d17480fcba37e572278fb4d14ec50f0bd. 1 of 1 blocking finding(s) are security-class and were withheld from adjudication, so the blocking verdict stands.

This comment is updated in place on each push.

BLOCKING -- website/src/pages/settings/DecisionsProviderPicker.tsx:443 -- Preserve the search-opened key field after highlight cleanup

selected ?? (openHostedKey && !blockedId(PRESET_JEV) ? PRESET_JEV : configuredChoice)
Trigger: With Laya, SageMaker, or “No model” configured and the hosted provider permitted, open the TypeSafe API key settings-search result. DecisionsCard supplies openHostedKey=true, then SettingsPage’s useSettingHighlight removes the highlight parameter after 100 ms.
Outcome: openHostedKey becomes false while selected remains null, restoring the configured option and unmounting the key field. The search destination disappears before the user can edit it.
Anchor: defect
Fix: Retain the search-opened entry in local state until the user selects another option, without writing a provider change.

[BLOCK-MERGE] 23655f1
[GPT-REVIEWED] 23655f1

Adjudication (Opus 5.5) — is blocking on each finding proportionate?

Harm is MEDIUM at most: a UI navigation regression with no security, credential or data harm. When the highlight param is removed (website/src/hooks/useSettingHighlight.ts:342/370/392), hostedKeyTargeted goes false (website/src/pages/settings/DecisionsCard.tsx:194-199), and chosen falls back to configuredChoice (website/src/pages/settings/DecisionsProviderPicker.tsx:443). Nothing gets written, and the user can recover by picking the hosted option by hand.

This finding was fenced, so FLAG is the only annotation available. The FLAG bar is "conditions so extreme a human would accept the residual risk", and that bar is not met: one settings-search click with a non-hosted provider configured triggers it on an ordinary path. The real fix is cheap: hold one piece of local state.

[ADJUDICATION] 23655f1d17480fcba37e572278fb4d14ec50f0bd total=0 uphold=0 downgrade=0
[GPT-ADJUDICATED] 23655f1d17480fcba37e572278fb4d14ec50f0bd

[ADJUDICATION-FENCED] 23655f1d17480fcba37e572278fb4d14ec50f0bd fenced=1 flagged=0
UPHOLD-FENCED F1 website/src/pages/settings/DecisionsProviderPicker.tsx:443 -- Ordinary settings-search use reaches it: highlight cleanup clears searchParams' `highlight`, which drops `openHostedKey` while `selected` is null, so the key field unmounts; no extreme condition is required, so a FLAG rationale does not hold.
[GPT-ADJUDICATED-FENCED] 23655f1d17480fcba37e572278fb4d14ec50f0bd

False positive or not applicable? A repository writer can comment:
/ai-review override gpt 23655f1d17480fcba37e572278fb4d14ec50f0bd false-positive: <reason, 20 to 500 characters>

@KiroCrewCaptain

Copy link
Copy Markdown
Collaborator

Code change approver recommendation: A person should decide

Why: the decider raised 1 possible problem and estimates a 50% chance that this change introduces a defect; a person should check it.

The per-model allow list is enforced consistently at every gateway enforcement point, fails closed if the policy changes mid-read, and leaves both the off switch and the standalone default (no policy: every model allowed) as they were. It should still not merge yet: a new test assigns .innerHTML, which breaks the repo's documented frontend rule and fails two blocking gates (Automated Rule Check and SAST). The fix is small. Many other checks were still pending at decision time.

Things to check (possible problems the decider raised; at this confidence many turn out to be fine):

  • The new self-check test builds its sample DOM with innerHTML. That breaks the repository's frontend rule and fails two blocking CI gates, so the PR cannot merge as it stands. Build the sample with createElement, textContent and setAttribute instead; the self-check still works and both gates should c (website/src/pages/settings/decisionModelNaming.test.tsx, medium)

An automated recommendation, not a review: a maintainer still decides. It approves on its own only when the estimated chance of a defect is at most 12% and the change has at most 400 changed lines. Decider v3c (217fe607) · commit 23655f1d17 · 4890 changed lines. If this recommendation is wrong, reply here: replies are how the approver is evaluated.

@github-actions github-actions Bot added the merge conflict Branch has merge conflicts with its base — author must resolve before merge label Nov 13, 2026
@bolichen97

Copy link
Copy Markdown
Contributor Author

Superseded: this diff was over the review lanes' 1 MB prefetch limit, so the work is split. #36271 carries the governance seam (the providers allow list on capabilities.decisions, gateway enforcement, provider_permitted, the disabled-by-administrator notice, the greyed "Not allowed by your administrator." states, the hosted provider's entry holding its API key) at about 227 KB, with screenshots attached instead of committed. The wording pass ("decision model" in every existing label, backend string and doc, the "Decision model chose" trace line, and the catalog and render tests that hold the UI to it) follows in a second PR opened on main once #36271 merges. Both findings from this run are fixed in the replacements: the autouse fixtures in test_decisions_provider_allowlist.py now use _floor_monkeypatch, and the naming test builds its sample with createElement instead of innerHTML.

@bolichen97 bolichen97 closed this Nov 13, 2026
auto-merge was automatically disabled November 13, 2026 01:36

Pull request was closed

@github-actions github-actions Bot removed the readiness: checking Automated validation is still running label Nov 13, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

merge conflict Branch has merge conflicts with its base — author must resolve before merge

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants