Repository navigation
Verify rendered dashboards and fix memory panel usability - #42
Merged
seathatflowsinourveins merged 3 commits intoSep 21, 2026
Merged
Conversation
|
You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard. |
seathatflowsinourveins
added a commit
that referenced
this pull request
Sep 23, 2026
…udged (Codex round 10) - claude-collect uses the repository the claude-args snapshot validated; an override or returned repo must name it exactly. - Each Codex adjudication worker rechecks its indexed input hash at start and after its calls. - inputs records each lane return's sha256; assemble refuses a layer whose lane return changed and writes lane_returns_sha256; record_verdicts.py refuses a new-wave adjudication that does not name the lane returns it seals. - Leak records keep input basenames only. - The layer-verdict workflow echoes args.launch (vendored from agent-lab #42, sha256 1a26ff79, registered); claude_lane.py requires that launch to name --repo and the tree's current digest instead of taking the launch digest from the caller. Tests: 6 new tests fail on the previous code; full suite 3610 OK (345 skipped). Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
seathatflowsinourveins
added a commit
that referenced
this pull request
Sep 23, 2026
…pository (agent-lab #42) Workflow sha256 35e793a0 replaces the interim 1a26ff79 on this branch (never on main); vendored-lanes.json, SHA256SUMS and the lane-provenance registry name it. Full suite 3610 OK (345 skipped). Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
seathatflowsinourveins
added a commit
that referenced
this pull request
Sep 24, 2026
…aths (Codex round 11) - record_verdicts.py rejects a Codex lane whose repo_tree_sha256 differs from the Claude lane's, and an adjudication whose provenance names another tree. - inputs reads each packet and lane return once (parse, hash and snapshot from the same bytes); record_verdicts.py reads each lane file once for validation, the lane_returns_sha256 check and sealing. - Windows host paths (drive, rooted backslash, UNC) are scrubbed, caught by the residual scan and redacted from leak text. - The Claude recipe passes --repo; vendored-lanes.json pins agent-lab #42's merge commit d47e8355. Tests: 5 new tests fail on the previous code; full suite 3615 OK (345 skipped). Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
seathatflowsinourveins
deleted the
codex/dashboard-render-e2e-20260921
branch
September 25, 2026 18:47
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
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.
The memory wiki appeared empty and the Grafana memory table was too wide to use. This change separates inventory, model health and provenance into readable panels, preserves an explicit unknown observation on failed captures, and corrects the guide: ai-memory's bundled wiki is read-only, System pages are collapsed, and browser search is global FTS.
The evidence covers rendered screenshots and native CLI/MCP/API comparisons for13 installed repository interfaces. It records the refreshed upstream SocratiCode graph and actual limitations: disabled memory consolidation, unavailable current Claude telemetry, optional FreeLLMAPI readiness warning, gateway-only zero compression savings, and historical Promptfoo fixtures. Private screenshots and returned files remain local; public projections retain hashes and upstream identities.
Validation:24 local adapter tests passed; publication validation passed for68 components,1299 hashed files and117 receipts; actual browser screenshots verified the repaired panels. Independent review checked491 retained bindings,48 viewer assets and13 screenshot hashes and resolved the failure-row and stale-screenshot findings. These checks are distinguished from upstream repository test suites and model-quality evaluation.