Repository navigation
feat(dashboard): drop the post-connect test run, explain failed runs on HQ - #1160
Merged
Merged
Conversation
…on HQ The owner decided the connect-check test run is not needed: the first real run already proves the credential. So: - No test after saving a credential. The Connect modal now says what was saved and that the next run uses it (plus the harness it switched to). No Test now / Test again, no /api/connect-check, no page polling hook. - The HQ Setup checklist marks "Model connected" done once a credential for the selected harness is saved. - ./aeon init drops step 7 "Test connection" and --no-test; Telegram is step 7 again. The sandbox test now asserts init never dispatches a run. - skills/connect-check and its aeon.yml entry are gone; catalog/skills.json, packs.json, skill-icons (+ dashboard icon map), eyebrowlock.json and the skill counts (86 -> 85) are back in step. The log reader is kept and repurposed: lib/connect-check.ts is now lib/run-diagnosis.ts. For a failed run it slices the Run output and error annotations and returns one plain reason, a next step, and whether the credential is the problem (and which harness ran, from the run banner). GET /api/runs/[id]/diagnosis reads the log with gh and caches the verdict. HQ Recent activity and the run detail panel show it under failed runs only, fetched lazily and cached, with a Connect button when the credential failed.
…out runs Match the hosted port (aeon-connect#60): - HQ Recent activity no longer reads the log of every failed row. Each failed or timed-out row has a "Why?" toggle; the log is read only when it is opened, and the per-run client cache is kept. The run detail panel still loads it right away, since that is one run the operator opened. - Runs that timed out are diagnosed too (server and client).
Match the hosted port: token expired / invalid_grant / revoked refresh token / expired session now read "The saved login has expired." with the next step "Log in again and paste the new login." (credential, so HQ offers Connect). A plain 401 or invalid key stays "The provider rejected the credential."
Copy the SIGNATURES list from aeon-connect#60 so canon and hosted say the same thing: "The saved login expired." / "Log in again and connect the new login." (also matches "refresh token expired"), and the hosted hints for rate limit, model not available and no usable credential.
A real Codex run whose ChatGPT login was revoked (aeon-oneshot-test run
37131474562) read as "The provider rejected the credential.": codex died
before the usage notice, so only the ##[error] annotation ("unauthorized
(401)") was read, never the harness output that says token_revoked.
- extractRunOutput: with no usage notice, anchor the slice on the first
##[error] line, so the failing step's own output is read.
- Expired-login signature also matches token_revoked and "invalidated
oauth token".
- Test built from the trimmed real log (gh shape, request ids and cf-ray
dropped): expired login, credential, harness codex.
…harness Same as hosted aeon-connect#60: when the "Using harness: X" banner is missing, the harness comes from the "effective model for X:" notice (outside script blocks). Banner wins when both are printed. Test copied from hosted.
5 tasks done
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What changed
/api/connect-check, no polling hook../aeon init: step 7 "Test connection" and--no-testare gone; Telegram is step 7 again.init-sandbox.shnow checks that init never dispatches a workflow run.skills/connect-checkand itsaeon.ymlentry.catalog/skills.json,catalog/packs.json, skill icons (+lib/skill-icons.data.ts),eyebrowlock.json, docs and skill counts (86 -> 85) updated.lib/connect-check.tsis nowlib/run-diagnosis.ts: for a failed run it reads the Run output and error lines and returns one plain reason, a next step, and whether the credential is the problem (for example "The provider rejected the credential. Next step: Paste a fresh key or log in again."). NewGET /api/runs/[id]/diagnosisreads the log withghand caches the result. HQ Recent activity and the run detail panel show it under failed runs only, loaded lazily and cached, with a Connect button when the credential failed.Verification
npm ci,npm run typecheck,npm run lint(0 errors),npm test(297 pass),npm run buildnpm run typecheck,npm run lint,bash apps/cli/test/init-sandbox.sh(PASS)bash scripts/tests/test_credential_manifest.sh(266 checks pass)bin/generate-skill-icons --check,validate-readme-catalog,validate-skill-packs,check-aeon-skill-sync,check-skill-categories,validate-config(+ its tests): all OK