Repository navigation
Preserve and audit credential registry refusal evidence - #156
Merged
Merged
Conversation
Preserve corrupt credential registries and emit actionable, secret-free diagnostics with durable refusal evidence across gateway startup, automatic provisioning, and offline auth commands. Keep audit sink failure beside the primary registry fault without changing exit or restart policy.\n\nThis repairs the unsafe diagnostic gap introduced by eac1ead (Commit credential lifecycle evidence with the registry state (#68)).\n\nAddresses #113
Translate unsafe registry files into the audited refusal path and publish refusal records only through trusted-root storage operations. Sync each audit directory parent before descending so a failed ancestry publication cannot be mistaken for durable evidence.\n\nAdd process regressions for unsafe files, symlink escape attempts, offline attribution, and the separate managed recovery record.
Create nested private directories beneath a trusted root one component at a time. Sync each established parent before descent and the final leaf before returning so audit sinks share one publication contract.\n\nUse the primitive for credential-registry refusal evidence and keep the sink's symlink escape regression.
Select one platform implementation for non-UTF registry paths so strict linting sees no unreachable fallback while preserving lossless audit targets.
Build adapter fixtures beneath an explicitly private auth directory so they exercise the same trusted-root contract as composition.
There was a problem hiding this comment.
Your trial has ended. Reactivate Greptile to resume code reviews.
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
Varuas37
pushed a commit
that referenced
this pull request
Nov 9, 2026
nessa_ui#156 merged as 2b5bba0, which holds the same tree as the branch head this PR was checked against (the last commit there only added a registry dependency). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01JcBpmVHPWxktA9QurvUMjM
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.
Corrupt or unsafe credential registries and their lock files now refuse admission with typed, secret-free diagnostics, exact target evidence, and role-specific recovery guidance. Ordinary startup, local provisioning, and offline auth preserve exit policy and audit the refusal without modifying rejected state. A lock refusal does not claim that an unread registry is corrupt or present. Audit delivery failures remain separate reported errors.
Private audit writes retain trusted-root authority through publication and cleanup. Temporary reservations acquire the Windows deletion capability they own; atomic no-replace publication preserves existing records. Directory guarantees distinguish Unix synchronization from Windows file flushing and write-through publication. Recovery guidance prefers restoring verified registry backups; lock repair never instructs registry reinitialization.
Relates to #113. This draft is not merge-cleared and does not close the issue. Round 3 findings led to a recorded structural diagnosis and revised ownership of cleanup capability, refusal transitions, and adversarial test setup. Final independent review remains pending.
Current revision validation is pending in the isolated local build queue and CI. It adds native Windows cleanup tests, valid/absent-registry unsafe-lock tests, and corrected ancestry-redirection tests with a complete malicious path. A controlled old-cleanup negative run must fail before the fixed-source validation is accepted.
Earlier revision validation (not proof for this new revision): strict all-target clippy for local-storage/auth/server; 9 local-storage tests, 62 auth unit and 3 integration tests, 8 real-process refusal tests, 32 server core tests, and architecture checks passed. Its CI passed seven checks; Windows exposed a path-display assertion mismatch corrected here while preserving exact typed target assertions.
Current formatting, architecture, and diff checks pass. Windows execution awaits CI. No power-loss testing or crash-durability proof is claimed. Shared journal-transaction APIs for the installer are a separate follow-up.