Add Claude Code GitHub Workflow

[wehappyfew]

🤖 Installing Claude Code GitHub App

This PR adds a GitHub Actions workflow that enables Claude Code integration in our repository.

What is Claude Code?

Claude Code is an AI coding agent that can help with:

• Bug fixes and improvements
• Documentation updates
• Implementing new features
• Code reviews and suggestions
• Writing tests
• And more!

How it works

Once this PR is merged, we'll be able to interact with Claude by mentioning @claude in a pull request or issue comment.
Once the workflow is triggered, Claude will analyze the comment and surrounding context, and execute on the request in a GitHub action.

Important Notes

• This workflow won't take effect until this PR is merged
• @claude mentions won't work until after the merge is complete
• The workflow runs automatically whenever Claude is mentioned in PR or issue comments
• Claude gets access to the entire PR or issue context including files, diffs, and previous comments

Security

• Our Anthropic API key is securely stored as a GitHub Actions secret
• Only users with write access to the repository can trigger the workflow
• All Claude runs are stored in the GitHub Actions run history
• Claude's default tools are limited to reading/writing files and interacting with our repo by creating comments, branches, and commits.
• We can add more allowed tools by adding them to the workflow file like:

allowed_tools: Bash(npm install),Bash(npm run build),Bash(npm run lint),Bash(npm run test)

There's more information in the Claude Code action repo.

After merging this PR, let's try mentioning @claude in a comment on any PR to get started!

[crucible-burnt]

Review — Claude Code GitHub Workflow

Identical to xion#483. Same notes apply — looks good, no blockers.

[crucible-burnt]

🔍 Crucible Security Review

Summary

Adds Claude Code GitHub Actions for the contracts repo — same pattern as xion repo.

Security Assessment

• Risk Level: Low ✅
• Identical configuration to burnt-labs/xion PR #483
• API key in secrets, appropriate permission scoping
• No write access to repository content

Immunefi Pattern Check

• N/A — CI/CD tooling

False Report Risk

• None

Code Quality Notes

• Consistent implementation across repos

Status

No security concerns. Standard AI-assisted review tooling.

[crucible-burnt]

🔍 Crucible Security Review

Summary

Adds two GitHub Actions workflows: Claude Code automated PR review and Claude Code interactive (via @claude mentions). CI/CD configuration only, no contract code changes.

Security Assessment

• Risk Level: Low

Findings:

1. Permissions scope (claude-code-review.yml): Uses id-token: write permission, which allows OIDC token generation. This is standard for Anthropic's action but worth noting — it means the workflow can request tokens that identify as the repo.

2. No write permissions for PRs: Both workflows only have pull-requests: read and issues: read. This means Claude cannot post comments or modify PRs. If the intent is for Claude to post review comments, pull-requests: write would be needed. As-is, the review output might only appear in workflow logs.

3. Secret exposure: CLAUDE_IN_GH_ACTIONS_API_TOKEN is properly stored as a GitHub secret. No hardcoded credentials.

4. Trigger scope (claude.yml): The interactive workflow triggers on issue_comment, pull_request_review_comment, issues, and pull_request_review. The @claude filter is good, but any user who can comment on issues/PRs can trigger the workflow and consume API credits.

Immunefi Pattern Check

• No contract code changes. No matches.

False Report Risk

• None. CI/CD only.

Code Quality Notes

• Consider adding pull-requests: write if Claude should actually post comments
• Consider restricting the @claude trigger to org members to prevent API credit abuse
• The plugin_marketplaces URL points to Anthropic's own repo — verify this is the intended source

Status

Low-risk CI addition. Minor suggestions on permissions and access control.

[crucible-burnt]

🔍 Crucible Security Review

Summary

Adds two GitHub Actions workflows for Claude Code integration — automated PR review and interactive @claude mentions. Low security surface.

Security Assessment

• Risk Level: Low

• Permissions are appropriately scoped: contents: read, pull-requests: read, issues: read, id-token: write (for OIDC)

• No write permissions to contents or pull-requests — Claude cannot modify code or approve PRs

• API key stored in secrets.CLAUDE_IN_GH_ACTIONS_API_TOKEN — standard GitHub secrets management

• The claude.yml workflow triggers on @claude mentions in issue/PR comments — this is public-facing and could be triggered by any commenter. Ensure the API key has appropriate rate limits/budgets to prevent cost abuse from external contributors spamming @claude

Immunefi Pattern Check

• N/A — CI/CD workflow only, no on-chain code changes

False Report Risk

• None

Code Quality Notes

• claude-code-review.yml uses fetch-depth: 1 — this is fine for review context but may limit Claude's ability to see full file history
• Consider adding a condition to claude.yml to restrict @claude to org members only (e.g., github.event.comment.author_association check) to prevent external abuse

Status

Clean from a security perspective. Minor recommendation to add author-association filtering on the interactive workflow.

[crucible-burnt]

🔍 Crucible Security Review

Summary

CI workflow addition for Claude Code automated reviews. No application code changes — purely DevOps/CI configuration.

Security Assessment

• Risk Level: Low
• Workflow permissions: Both workflows request contents: read, pull-requests: read, issues: read, id-token: write. The id-token: write permission is required by the Anthropic action for OIDC token exchange — standard pattern, not a concern.
• Secret exposure: Uses secrets.CLAUDE_IN_GH_ACTIONS_API_TOKEN — ensure this is a dedicated API key with appropriate rate limits, not a shared org-wide key.
• Trigger scope: claude.yml triggers on @claude mentions in comments/issues/reviews. This is open to any commenter — verify the Anthropic action itself gates who can invoke it (e.g., org membership check), otherwise any GitHub user could trigger API calls by commenting @claude on public issues.
• actions: read in additional_permissions: Low risk, allows Claude to read CI results.

Immunefi Pattern Check

• No matches — this is CI configuration only, no smart contract or chain logic changes.

False Report Risk

• None — no application code involved.

Code Quality Notes

• Consider adding path filters to claude-code-review.yml to avoid triggering on docs-only PRs (commented out currently).
• The claude.yml workflow has no author filtering enabled (all commented out). For a public repo, this means anyone can trigger Claude by commenting. Recommend enabling the author_association filter or a team-based allowlist.
• Both workflows use fetch-depth: 1 which is fine for review context.

Status

Low-risk CI addition. Main recommendation: enable author filtering on claude.yml to prevent abuse on public repos. No security blockers.

[crucible-burnt]

🔍 Crucible Security Review

Summary

Adds two GitHub Actions workflows: Claude Code Review (auto-triggered on PRs) and Claude Code (triggered by @claude mentions in issues/PRs). CI config only.

Security Assessment

• Risk Level: Low
• Permissions scoping is appropriately minimal: contents: read, pull-requests: read, issues: read, id-token: write. No write permissions to contents or PRs — Claude cannot merge or modify code.
• Secret reference: ${{ secrets.CLAUDE_IN_GH_ACTIONS_API_TOKEN }} — ensure this secret is scoped to this repo only and rotated periodically.
• id-token: write: Required for OIDC authentication by the action. Standard for Anthropic's action.
• actions: read in claude.yml as additional permission — allows Claude to read CI results. Acceptable.
• Third-party action: anthropics/claude-code-action@v1 — pinned to v1 tag (major version). Consider pinning to a specific SHA for supply chain security: anthropics/claude-code-action@<sha>.
• Plugin marketplace: claude-code-review.yml uses plugin_marketplaces and plugins — the marketplace URL points to the official anthropic repo, which is fine, but plugins could be a vector if the marketplace is compromised.

Immunefi Pattern Check

• Not applicable — CI config only.

False Report Risk

• None.

Code Quality Notes

• Clean workflow configuration.
• Good separation: one workflow for automated review on all PRs, another for interactive Claude usage via @mentions.
• Consider adding path filters to claude-code-review.yml to avoid running on docs-only PRs.

Status

Clean. Minor recommendation to pin the action to a commit SHA for supply chain hardening.