Skip to content

security auth write deny - #70962

Open
webtecnica wants to merge 1 commit into
NousResearch:mainfrom
webtecnica:fix/security-auth-write-deny
Open

security auth write deny#70962
webtecnica wants to merge 1 commit into
NousResearch:mainfrom
webtecnica:fix/security-auth-write-deny

Conversation

@webtecnica

Copy link
Copy Markdown
Contributor

Fixes issue

@alt-glitch alt-glitch added type/security Security vulnerability or hardening P3 Low — cosmetic, nice to have comp/agent Core agent runtime: loop, agent_init, prompt builder, context-compression, responses endpoint tool/file File tools (read, write, patch, search) area/auth Authentication, OAuth, credential pools duplicate This issue or pull request already exists labels Jul 24, 2026
@alt-glitch

Copy link
Copy Markdown
Collaborator

This was generated by AI during triage.

Duplicate of #70967. It contains the same auth.json/auth.lock write-deny mechanism and additionally covers the related Google OAuth and webhook credential stores.

@teknium1

Copy link
Copy Markdown
Contributor

Thanks for addressing a real credential-store protection gap. Current main confirms that auth.json is read-denied in agent/file_safety.py:274-280 but is absent from the write-deny set consumed by file operations and ACP writes (agent/file_safety.py:106, tools/file_operations.py:1293, agent/copilot_acp_client.py:735).

Problems

  • The patch is partial: the same read-side credential list also contains webhook_subscriptions.json, auth/google_oauth.json, and cache/bws_cache.json (agent/file_safety.py:279-285), yet the proposed write list adds only auth.json and auth.lock. Those sibling stores remain writable by the same mechanism.
  • The added tests cover the predicate only. A representative production-path test would better validate the security boundary, since deletion, moves, writes, patches, and ACP writes all consume the shared guard.

Suggested changes

  • Extend the write-deny entries for the remaining read-denied credential stores at both active-profile and root locations.
  • Add a representative file-operation or ACP write-denial test.

This is an automated hermes-sweeper review.

@teknium1 teknium1 added sweeper:risk-security-boundary Sweeper risk: may affect sandboxing, auth, credentials, or sensitive data sweeper:blast-moderate Sweeper blast radius: moderate — a subsystem or single platform labels Jul 30, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

area/auth Authentication, OAuth, credential pools comp/agent Core agent runtime: loop, agent_init, prompt builder, context-compression, responses endpoint duplicate This issue or pull request already exists P3 Low — cosmetic, nice to have sweeper:blast-moderate Sweeper blast radius: moderate — a subsystem or single platform sweeper:risk-security-boundary Sweeper risk: may affect sandboxing, auth, credentials, or sensitive data tool/file File tools (read, write, patch, search) type/security Security vulnerability or hardening

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants