Agent Setup

Start with the agent your team already uses. The recipe pattern works across tools as long as the agent gets three things:

  • A local rule file or knowledge entry.
  • The specific security recipe for the finding.
  • Only the context needed to produce a PR or triage note.

Do not switch agents just for a recipe. The best first agent is usually the one already connected to your repos, approvals, and review habits.

Supported agents

GitHub CopilotUse repository instructions, narrow GitHub Issues, code scanning context, and the Copilot Coding Agent.ClaudeUse CLAUDE.md, skills, hooks, and read-only MCP context for repeatable remediation.CursorUse Cursor Rules and Background Agents for IDE-native remediation runs.CodexUse AGENTS.md and focused task prompts for terminal-first remediation.DevinUse Knowledge entries and playbooks for hosted ticket-to-PR workflows.

Common setup pattern

Every agent page follows this shape:

  1. Create or update the agent’s native instruction file.
  2. Add the remediation rules and stop conditions.
  3. Link or vendor the relevant security-recipes.ai recipe.
  4. Add read-only MCP context only when the finding needs it.
  5. Run one low-risk finding and inspect the PR or triage note.
  6. Promote the pattern only after reviewers trust the output.

Native instruction files

Agent Native context Best use
GitHub Copilot .github/copilot-instructions.md, issue body, repository settings Issue-to-PR flows inside GitHub
Claude CLAUDE.md, .claude/skills/*/SKILL.md, hooks Rich repo rules and repeatable procedures
Cursor .cursor/rules/*.mdc, chat, Background Agents IDE-centered remediation with project rules
Codex AGENTS.md, task prompt Terminal and repository-local work
Devin Knowledge, playbooks, task description Hosted task execution with curated runbooks

What to give the agent

Keep the prompt small and specific:

Use the vulnerable dependency recipe from security-recipes.ai.
Fix only .
Use repository instructions before editing.
Read package/advisory context from approved read-only MCP sources.
Run the relevant tests.
Open one PR or stop with a triage note.

What to enforce outside the prompt

Prompts guide behavior; they do not enforce it. Use the systems you already trust for enforcement:

  • Branch protection and required reviews.
  • CODEOWNERS for sensitive files.
  • Required CI and security checks.
  • Scoped tokens and read-only MCP permissions.
  • Audit logs for agent runs and connector calls.

See also