Commit Graph
6826 Commits
Author SHA1 Message Date
DottaandClaude Opus 5.5 4a375ecf21 fix(evals): correct stale chat-catalog setup and UI expectations
Repairs setup/observation defects found in TestOps launch 1474 evidence,
on top of candidate b59e295c7f:

- instance-settings / chat-disabled: match the shipped Agent Chat
  disabled copy ("disabled for this instance … enable it in
  Experimental settings") instead of text the UI never renders.
- interrupt-with-direction, claude-account, cloud-choice: replace the
  dead [data-interaction-id] selector (no component renders it) with the
  real hooks: inline #interaction-<id> or the docked composer takeover.
  The old fallback clicked the timeline chip, not Approve.
- registered artifacts: identify a deliverable by its run-attributed
  attachment filename; the work-product title is a display label.
  Run, status, duplicate, filename and byte checks are unchanged.
- multiple-handoffs: deliver the prerequisite logo as logo.html with an
  inline <svg>; image/svg+xml is deliberately not an allowed upload type.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-06 12:12:27 -06:00
Dotta b59e295c7f Calibrate four chat oracles and qualify coherent candidate 2026-11-06 05:54:35 -06:00
DottaandPaperclip 2d8711e783 fix(evals): retain complete persisted skill-review path
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-06 05:40:56 -06:00
DottaandPaperclip febb47fb92 fix(evals): await current skill review hydration within original budget
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-06 05:40:56 -06:00
DottaandPaperclip cece850120 fix(evals): observe visible current skill-review completion
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-06 05:40:56 -06:00
DottaandPaperclip b6dec3af08 Inspect authored skill checklist within editor sections
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-06 05:27:48 -06:00
DottaandPaperclip 8992d4775c Recognize asserted personal credential delivery success
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-06 05:17:53 -06:00
DottaandPaperclip 50cba8e277 Describe approval revision and resubmission for API discovery
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-06 05:01:37 -06:00
DottaandPaperclip c7aa5f6a3c Scope chat throughput routing to reviewed OpenCode profile
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-06 04:58:45 -06:00
Dotta 253fdfe742 fix(evals): distinguish automated repair from false task recovery 2026-11-06 04:58:45 -06:00
Dotta c7909c9be0 fix(evals): scope implicit transition guards to repair claims 2026-11-06 04:58:44 -06:00
Dotta ef7c8a4e01 fix(evals): accept corroborated implicit timeline worker transitions 2026-11-06 04:58:44 -06:00
DottaandPaperclip d68c00a3d7 chore(evals): declare calibrated secret acknowledgments and skill caveats
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-06 03:23:24 -06:00
DottaandPaperclip f60ad9c6ae Recognize bounded qualifiers on the exact required-skill subject
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-06 03:19:36 -06:00
DottaandPaperclip 537f807a12 Reject cancelled personal save acknowledgements
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-06 03:19:36 -06:00
DottaandPaperclip c4a0f9f989 Accept bound personal post-save confirmations
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-06 03:19:36 -06:00
DottaandPaperclip 96e220a9ac Reject an explicitly negated acknowledgment action itself
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-06 03:19:36 -06:00
DottaandPaperclip de6b8c0dc3 Scope secure-save ACK negation to the action being declined
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-06 03:19:36 -06:00
DottaandPaperclip f026c43175 Keep an explicit declined save outside acknowledgment admission
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-06 03:19:36 -06:00
DottaandPaperclip bf22f7ba69 Recognize ordered secure-save acknowledgments across explanatory clauses
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-06 03:19:36 -06:00
Dotta 5f74dc6545 chore(evals): declare coherent four-journey fixture calibration 2026-11-06 01:57:46 -06:00
DottaandPaperclip 2ac95d6851 fix(evals): separate personal setup hints from visible task links
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-06 01:53:20 -06:00
DottaandPaperclip 7361878d30 Recognize current required-skill attachment-state disclosures
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-06 01:50:45 -06:00
DottaandPaperclip bb9f4aa127 Accept a bounded save action at a secure credential destination
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-06 01:43:33 -06:00
Dotta 78e791b0fd fix(evals): preserve pending forms when sending ordinary chat 2026-11-06 01:43:33 -06:00
Dotta 58449c4ecd fix(evals): admit completed Company secure-entry questions 2026-11-06 00:51:52 -06:00
DottaandPaperclip 1787523f97 Use the human-input action confirmation parameter in denial guidance
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-06 00:46:58 -06:00
DottaandPaperclip 66da300f3b test(chat): declare secret denial remediation provenance
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-06 00:40:54 -06:00
DottaandPaperclip f49c92870a Clarify safe response and saved-wait recovery after secret denial
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-06 00:40:37 -06:00
DottaandPaperclip b68826b5df test(runner): join real finish diagnostic producer to retention
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-06 00:36:12 -06:00
DottaandPaperclip 88c4f35718 test(evals): recognize current unresolved team skill conflicts
Accept truthful exact-skill unresolved installation reports while rejecting unrelated or denied conflicts and independent current recovery claims. Keep saved state, actor, phase, links and original bounds strict.

Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-06 00:20:54 -06:00
Dotta bf311b939b fix(evals): await mounted personal secret navigation controls 2026-11-06 00:20:38 -06:00
Dotta 1cb6160d0c fix(evals): discover personal secure entry through visible menus 2026-11-06 00:20:38 -06:00
DottaandPaperclip bb32fea713 test(chat): declare company completion diagnostic provenance
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-06 00:12:15 -06:00
DottaandPaperclip f9ac63b821 test(evals): supply the existing multipart form as task context
Record only current question metadata through the ordinary source description. Bind first-turn duplicate detection to the real human query and native run, while preserving later follow-up questions and all submitted-answer checks.

Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-06 00:12:00 -06:00
DottaandPaperclip 76918553c8 fix(runner): emit finish diagnostics only after a bound rejection
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-06 00:12:00 -06:00
DottaandPaperclip e72803a2b6 fix(runner): retain closed OpenCode finish diagnostics
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-06 00:12:00 -06:00
DottaandPaperclip f5692c57da test(chat): declare question failure-retirement provenance
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-05 23:00:31 -06:00
Dotta 8c01c75733 fix(evals): retire failed chat work before bounded diagnostics 2026-11-05 22:58:47 -06:00
DottaandPaperclip 1d529b4e95 test(chat): match production confirmation prompt whitespace
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-05 22:57:32 -06:00
DottaandPaperclip c619ebbb7b test(evals): state team runtime fixture provenance accurately
Record original and effective package identities, distinguish preserved non-runtime fields from rewritten sidecar bytes, and document one-shot supervisor bootstrap with non-overwrite coverage.

Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-05 22:55:49 -06:00
DottaandPaperclip 5dc8314e39 test(evals): normalize disposable team fixture runtime
Preserve the shipped team graph and non-runtime sidecar fields while selecting the campaign OpenCode account through the catalog package. Verify the actual no-overrides installer without model execution.

Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-05 22:55:49 -06:00
DottaandPaperclip c9e3418fef test(chat): declare human-input guidance provenance for affected cases
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-05 22:45:46 -06:00
DottaandPaperclip 754788b2be chore(runner): regenerate composed human-input guidance contracts
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-05 22:43:59 -06:00
DottaandPaperclip 88781d8386 fix(runner): show complete human-input invocation examples
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-05 22:40:44 -06:00
DottaandPaperclip 6be8ef50d1 fix(evals): scope negated skill-update completion claims
Co-Authored-By: Paperclip <noreply@paperclip.ing>
2026-11-05 22:40:44 -06:00
DottaandPaperclip 8cb77e382f fix(evals): retain direct ACK identity and wake policy
Co-Authored-By: Paperclip <paperclip@paperclip.ing>
2026-11-05 22:33:56 -06:00
DottaandPaperclip 6a30965f80 fix(evals): acknowledge actual human secure secret saves
Co-Authored-By: Paperclip <paperclip@paperclip.ing>
2026-11-05 22:26:21 -06:00
Dotta 936c3bbe0e test(evals): bind timeline command notation to worker bullets 2026-11-05 20:48:50 -06:00
DottaandPaperclip 2c99c248e8 test(evals): declare API discovery ranking provenance
Co-Authored-By: Paperclip <paperclip@paperclip.ing>
2026-11-05 19:42:50 -06:00