fix(providers): correct Gitlawb GMI probing and drain stale baseline by bloodf · Pull Request #124 · bloodf/durindoor

bloodf · 2026-07-09T00:17:52Z

Scope

This reworks the Gitlawb GMI connection probe on current dev@dd11086b69b835fd80a2553ecda6169d2ffa9bf5 and consolidates the 16 verified stale-baseline deletions from #118, #119, #123, and #124.

Correct Gitlawb GMI runtime/probe traffic to https://opengateway.gitlawb.com/v1/gmi-cloud/chat/completions.
Keep model discovery on https://opengateway.gitlawb.com/v1/gmi-cloud/models and publish an OpenAI-compatible model fetcher.
Resolve the registry provider to its configured alias so the probe sends the real default model (XiaomiMiMo/MiMo-V2.5-Pro).
Treat 2xx/400/422/429 as authenticated, 401/403 as credential rejection, 404 as an endpoint error, and 5xx/network failures as provider unavailability.
Propagate sanitized probe failures through both connection-test and provider-validation APIs without exposing API keys.
Replace the MITM test's inline fake with isolated calls to production ensureRootCASync.
Register translators explicitly in the normalization test and refresh the Gitlawb golden URL and generated agent index.

Source and intentional divergence

The original OmniRoute provider source is commit 3ddcee6369c54e1c844a6e46cbbc79870d10d30b. The relevant registry, validator, and test blobs are unchanged on audited OmniRoute main@9cd18bf9a11b7d2e8c037c374631b492adabf469. The detailed source lineage is OmniRoute #2837.

OmniRoute #2476 later made Gitlawb keys optional. DurinDoor still requires a key in its API-key UI and validation route. This PR deliberately defers optional-key parity until UI, API, no-auth runtime, persistence/import, and end-to-end tests can be changed together.

Baseline impact

tests/__baseline__/known-fails.txt changes from 67 to 51 entries: exactly 16 deletions and zero additions.

test(baseline): drain mitm-rootca* regressions from 3-day window #118: 6 production MITM Root CA cases.
test(baseline): drain translator-request-normalization regressions from 3-day window #119: 4 translator request-normalization cases.
test(baseline): drain xai-oauth-service regressions from 3-day window #123: 5 xAI OAuth service cases.
fix(providers): correct Gitlawb GMI probing and drain stale baseline #124: 1 Gitlawb GMI connection-probe case.

The isolated local fail-closed gate on Node 20.20.2/npm 10.8.2 reports 17 raw failures, all 17 known, 34 stale entries, zero additions, and no startup/collection error. Linux CI remains authoritative before merge.

Verification

All commands used temporary HOME and DATA_DIR values.

# clean exact dependency installations
npm ci
cd tests && npm ci

# focused behavior and regression coverage
npx vitest run --config vitest.config.js \
  unit/gitlawb-gmi-connection-test.test.js \
  unit/provider-test-utils.test.js \
  unit/provider-validation.test.js \
  translator/golden-url-header.test.js \
  unit/omniroute-missing-api-cloud-providers.test.js \
  unit/mitm-root-ca.test.js \
  unit/mitm-rootca-autogen.test.js \
  unit/translator-request-normalization.test.js \
  unit/xai-oauth-service.test.js
# 9 files, 402 tests passed

BASELINE_BASE_REF=origin/dev npm run test:ci
# exit 0; raw 17, known 17, stale 34, additions 0

npm run lint -- --quiet
npm run check:agent-index
npm run build
npx --package=@commitlint/cli@20.5.3 \
  --package=@commitlint/config-conventional@20.5.0 \
  commitlint --config .commitlintrc.cjs --from=origin/dev --to=HEAD
# all exit 0

Fresh JSON and JUnit reports were generated and parsed locally; CI uploads both artifacts on every run.

Documentation, migration, and compatibility

Updated scoped OmniRoute review documentation and regenerated open-sse/AGENT-INDEX.md.
No schema migration or stored-data rewrite.
No API-key secret rotation, rewrite, or logging.
No change to legacy key shapes, provider aliases, request headers, or accepted data paths.
No root REPORT.md, dependency symlink, or baseline addition is included.

Merge policy: squash only after all required checks pass and all review threads are resolved.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 3571fb130b

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

Continue draining the known-fails.txt entries that TeamBaselineOtherV2 identified in the 3-day review. This batch covers: - codex-refresh-token (2) - force-stream-config (2) - db-driver-chain (1) - db-sqlite-vs-lowdb (1) — source fix in src/lib/db/repos/usageRepo.js - build-models-list-noauth (1) — source fix in src/app/api/v1/models/buildModelsList.js - command-code-validation (1) — source fix in src/app/api/providers/validate/route.js - gitlab-duo-registry (1) — source fix in open-sse/providers/registry/gitlab-duo.js - omniroute-simple-a-providers (2) — source fix in open-sse/providers/registry/agentrouter.js - omniroute-simple-a-review-fixes (2) — source fix in src/app/api/providers/validate/route.js + src/app/api/providers/suggested-models/filters.js - omniroute-simple-c-providers (1) — source fix in open-sse/providers/registry/index.js - zenmux-free (1) — source fix in open-sse/executors/index.js - openai-to-kiro (1) - agentrouter-provider (2, collateral) The 3 entries already in PR #124 (gitlawb-gmi, providerProbe, one known-fails line) are NOT touched here. This branch is a clean rebase onto origin/dev with only the additional changes from the 9-file batch.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 5584f8f779

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 74cd58b300

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

…piKeyUsageTotals-dependent entries)

…awb-gmi JSDoc) - open-sse/providers/registry/gitlawb-gmi.js: JSDoc on the probeUsesBaseUrl flag (Codex comment id 3548101977). - src/app/api/providers/providerProbe.js: resolve provider alias before getDefaultModel so the probed model comes from the registry (XiaomiMiMo/MiMo-V2.5-Pro) instead of falling back to 'test' (Codex comment id 3548048472). - tests/__baseline__/known-fails.txt: remove the gitlawb-gmi entry now that the test exercises the base URL with a real model and succeeds (Codex comment id 3548083460). - tests/unit/gitlawb-gmi-connection-test.test.js: assert the body uses the registry's MiMo-V2.5-Pro model id. Recreated from inspected commit 809b9819a; manually committed under durindoor-agent identity to satisfy user model constraint (no GPT-5.5).

chatgpt-codex-connector Bot reviewed Jul 9, 2026

View reviewed changes

Comment thread src/app/api/providers/providerProbe.js Outdated

bloodf force-pushed the fix/v2/baseline-other branch from 3571fb1 to 5584f8f Compare July 9, 2026 00:31

bloodf changed the title ~~test(baseline): drain gitlawb-gmi connection test from 3-day window~~ fix(providers): gitlawb-gmi validateUrl + providerProbe baseUrl-only fallback (baseline test entry retained) Jul 9, 2026

bloodf mentioned this pull request Jul 9, 2026

test(baseline): drain remaining 3-day window regressions (batch 2) #126

Closed

chatgpt-codex-connector Bot reviewed Jul 9, 2026

View reviewed changes

Comment thread tests/__baseline__/known-fails.txt Outdated

bloodf force-pushed the fix/v2/baseline-other branch 3 times, most recently from e1ef7eb to 74cd58b Compare July 9, 2026 00:37

chatgpt-codex-connector Bot reviewed Jul 9, 2026

View reviewed changes

Comment thread tests/node_modules Outdated

Comment thread open-sse/providers/registry/gitlawb-gmi.js

test(baseline): drain remaining 3-day window regressions (excluding a…

852a8d5

…piKeyUsageTotals-dependent entries)

bloodf force-pushed the fix/v2/baseline-other branch from 74cd58b to 852a8d5 Compare July 9, 2026 01:07

bloodf added 2 commits July 10, 2026 03:04

Merge remote-tracking branch 'origin/dev' into fix/v2/baseline-other

f61b3c4

fix(providers): correct Gitlawb GMI probing and drain stale baseline

0539d58

bloodf changed the title ~~fix(providers): gitlawb-gmi validateUrl + providerProbe baseUrl-only fallback (baseline test entry retained)~~ fix(providers): correct Gitlawb GMI probing and drain stale baseline Jul 10, 2026

bloodf merged commit 6457d95 into dev Jul 10, 2026
3 checks passed

bloodf deleted the fix/v2/baseline-other branch July 17, 2026 02:49

bloodf mentioned this pull request Jul 20, 2026

merge: DurinDoor 2.2.0 release #360

Merged

7 of 8 tasks

Navigation Menu

Repository navigation

fix(providers): correct Gitlawb GMI probing and drain stale baseline - #124

fix(providers): correct Gitlawb GMI probing and drain stale baseline#124
bloodf merged 4 commits into
devfrom
fix/v2/baseline-other

bloodf commented Jul 9, 2026 •

edited

Loading

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

bloodf commented Jul 9, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Scope

Source and intentional divergence

Baseline impact

Verification

Documentation, migration, and compatibility

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

bloodf commented Jul 9, 2026 •

edited

Loading