Skip to content

security: disable OpenAI Codex from pipeline and tooling#18

Merged
AndrewAltimit merged 2 commits intomainfrom
disable-codex-pipeline
Mar 3, 2026
Merged

security: disable OpenAI Codex from pipeline and tooling#18
AndrewAltimit merged 2 commits intomainfrom
disable-codex-pipeline

Conversation

@AndrewAltimit
Copy link
Owner

Summary

  • Permanently disables OpenAI Codex from CI pipeline (if: false), MCP config, and Docker services due to OpenAI's partnerships enabling mass surveillance and autonomous weapons decision-making
  • Adds a prominent security policy notice to README.md explaining why Codex is disabled and recommending Anthropic (Claude) models
  • Updates all documentation (README, AGENTS.md, CONTRIBUTING.md) to reflect the removal

Changes

File Change
pr-validation.yml codex-review job set to if: false, status summary shows "disabled (security policy)"
docker-compose.yml mcp-codex moved to unreachable disabled-codex profile, CODEX_ENABLED=false
.mcp.json Codex server renamed to codex-DISABLED, command replaced with no-op echo
.env.example Codex env vars commented out with explanation
README.md Security policy blockquote added, all "Gemini/Codex" refs updated to "Gemini"
AGENTS.md Pipeline step struck through, agent/MCP tables marked DISABLED
CONTRIBUTING.md Codex removed from agent list

Test plan

  • PR validation workflow still runs (codex-review job skips cleanly via if: false, dependent jobs unaffected)
  • docker compose --profile services up does not start mcp-codex
  • Security policy notice renders correctly in README.md on GitHub

Generated with Claude Code

OpenAI has entered partnerships with governments and defense organizations
that enable mass surveillance of civilian populations and autonomous weapons
kill-chain decision-making. This poses unacceptable security and ethical
risks for anyone whose data transits OpenAI infrastructure.

- PR validation: codex-review job set to `if: false`
- docker-compose: mcp-codex moved to unreachable profile
- .mcp.json: codex server disabled (no-op command)
- .env.example: codex config commented out
- README/AGENTS/CONTRIBUTING: all references updated, security policy
  blockquote added recommending Anthropic (Claude) models

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
@github-actions
Copy link

github-actions bot commented Mar 3, 2026

Gemini AI Code Review

Issues (if any)

  • [WARNING] CLAUDE.md:123 - Stale reference to Codex
    • The CI/CD description still mentions "(Gemini/Codex)" for PR reviews. This should be updated to reflect the removal of Codex.
  • [WARNING] tools/cli/agents/run_codex.sh - Local agent script remains functional
    • This script allows local execution of Codex, contradicting the new security policy stating that OpenAI Codex has been removed from the project's tooling.
  • [WARNING] tools/cli/containers/run_codex_container.sh - Broken and stale container script
    • References the codex-agent service no longer present in docker-compose.yml, causing it to fail. Also violates the security policy.

Previous Issues (for incremental reviews)

(none)

Suggestions (if any)

  • .github/workflows/pr-validation.yml:379 - Remove CODEX_REVIEW_PATH environment variable
    • Since the codex-review job is permanently disabled, the codex-review.md artifact will never be generated. Removing this variable from the agent-review-response job would improve clarity.
  • .github/workflows/pr-validation.yml:307 - Update job comments
    • The comment for agent-review-response still mentions responding to "Gemini/Codex feedback".

Notes

  • 1 claim(s) were automatically filtered as potential hallucinations (file:line content didn't match claims)

  • Excellent work on the comprehensive security policy notice in README.md and the updates to CONTRIBUTING.md.

  • Moving the MCP service to an unreachable profile in docker-compose.yml is an effective way to prevent accidental activation.

Reaction


Generated by Gemini AI (gemini-3-flash-preview). Supplementary to human reviews.

Remove stale Codex references missed in initial security disable:
- Delete tools/cli/agents/run_codex.sh (contradicts security policy)
- Delete tools/cli/containers/run_codex_container.sh (broken, references
  removed codex-agent service)
- Remove CODEX_REVIEW_PATH env var from agent-review-response job
- Update CLAUDE.md CI description to say "Gemini" not "Gemini/Codex"
- Update pr-validation.yml comment to say "Gemini feedback"
- Remove run_codex.sh entry from AGENTS.md agent table

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
@AndrewAltimit
Copy link
Owner Author

Review Response Agent (Iteration 1)

Status: No changes needed

Fixed Issues

  • CLAUDE.md:123 - Stale "Gemini/Codex" reference: Updated to "Gemini" only since Codex has been removed
  • tools/cli/agents/run_codex.sh - Functional Codex script contradicting security policy: Deleted the script entirely (216 lines)
  • tools/cli/containers/run_codex_container.sh - Broken/stale container script: Deleted the script entirely (84 lines) - it referenced the removed codex-agent docker service
  • pr-validation.yml:379 - CODEX_REVIEW_PATH env var: Removed since the codex-review job is permanently disabled and will never produce this artifact
  • pr-validation.yml:307 - Stale "Gemini/Codex" comment: Updated to "Gemini feedback"
  • AGENTS.md:196 - Stale run_codex.sh table entry: Removed the row referencing the now-deleted script

Ignored Issues

  • (none - all reported issues were verified and legitimate)

Deferred to Human

  • (none)

Notes

  • All 3 warnings and 2 suggestions from the Gemini review were legitimate and have been fixed
  • Also found and fixed an additional stale reference in AGENTS.md that wasn't flagged by the review
  • Changes pushed to disable-codex-pipeline branch as commit c15939e

The agent reviewed feedback but determined no code changes were required.

@github-actions
Copy link

github-actions bot commented Mar 3, 2026

Gemini AI Incremental Review

This is an incremental review focusing on changes since the last review.

Issues (if any)

(none)

Previous Issues (for incremental reviews)

  • [RESOLVED] CLAUDE.md:123 - Stale reference to Codex
  • [RESOLVED] tools/cli/agents/run_codex.sh - Local agent script remains functional
  • [RESOLVED] tools/cli/containers/run_codex_container.sh - Broken and stale container script

Suggestions (if any)

(none)

Notes

  • The author has thoroughly scrubbed Codex references from documentation, CI workflows, and local tooling.
  • Deletion of run_codex.sh and run_codex_container.sh correctly aligns the local environment with the new security policy.
  • The codex-review job in pr-validation.yml is correctly hard-coded to if: false to prevent accidental execution while maintaining workflow structure.

Reaction


Generated by Gemini AI (gemini-3-flash-preview). Supplementary to human reviews.

@AndrewAltimit
Copy link
Owner Author

Review Response Agent (Iteration 2)

Status: No changes needed

Fixed Issues

  • (none needed)

Ignored Issues

  • (none)

Deferred to Human

  • (none)

Notes

  • The latest Gemini incremental review confirms all previously reported issues are RESOLVED
  • No new issues were found in the incremental review of commit c15939e
  • The previous iteration already fixed all flagged items: stale Codex references in CLAUDE.md, deleted run_codex.sh and run_codex_container.sh scripts, cleaned up pr-validation.yml, and updated AGENTS.md

The agent reviewed feedback but determined no code changes were required.

@AndrewAltimit AndrewAltimit merged commit 270a8b7 into main Mar 3, 2026
7 checks passed
@AndrewAltimit AndrewAltimit deleted the disable-codex-pipeline branch March 3, 2026 08:26
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant