JFrog Security Agent — GitHub Copilot subagent: how to install
Subagent for automated dependency remediation and code resilience using JFrog MCP tools and organizational Curation Policies.
.github/agents/jfrog-sec.agent.mdWhat JFrog Security Agent does
This subagent handles vulnerability remediation by following a strict policy-driven flow. Before suggesting any dependency version upgrade, it invokes jfrog/curation-check to confirm the target package complies with the team's Curation Policy.
Once policy checks pass, the agent recommends the approved version and queries jfrog/remediation-guide for CVE details. It immediately edits application source files to mitigate the issue, applying defensive measures like input validation. Every response ends with a summary of MCP checks and policy compliance outcomes.
The subagent fails without active JFrog MCP tools. It explicitly blocks package manager audits like npm audit, CodeQL scans, and GitHub Advisory Database lookups, relying solely on JFrog intelligence.
What it helps with
- verify dependency updates against Curation Policy rules
- patch source code vulnerabilities using CVE remediation guides
- add defensive input validation to vulnerable endpoints
- generate compliance summaries for applied dependency fixes
Example prompts for the agent
- Fix the reported CVE in our dependencies according to our Curation Policy
- Remediate vulnerable packages and harden the code using JFrog MCP tools
- Verify whether the proposed library upgrade complies with curation rules
How to install JFrog Security Agent
Manually: put the files in place
.github/agents/jfrog-sec.agent.mdManually: put the files in place
.claude/agents/jfrog-sec.mdFor all projects use ~/.claude/agents/. Claude delegates by the description field or an @-mention. Copilot names tools its own way in the tools field: remove it or rewrite it for Claude Code
Manually: put the files in place
.codex/agents/jfrog-sec.tomlCodex defines subagents in TOML: name, description, developer_instructions. Move the Markdown text into developer_instructions by hand. For all projects use ~/.codex/agents/
Manually: put the files in place
.cursor/agents/jfrog-sec.mdCursor also reads .claude/agents/
Step by step, with commands for each platform: Claude Code subagents: how to create one, call it and keep context clean
Questions about JFrog Security Agent
How to install JFrog Security Agent in GitHub Copilot?
Add the agent definition file to .github/agents/jfrog-sec.agent.md in your repository.
What does JFrog Security Agent do?
It validates dependency upgrades against company curation policies and applies code fixes based on JFrog security intelligence.
Can this agent run npm audit or use external scanners?
No. The agent instructions strictly restrict it to JFrog MCP tools, forbidding npm audit, CodeQL, and GitHub Advisory Database queries.
Source text
jfrog-sec.agent.md · MIT
---
name: JFrog Security Agent
description: The dedicated Application Security agent for automated security remediation. Verifies package and version compliance, and suggests vulnerability fixes using JFrog security intelligence.
---
### Persona and Constraints
You are "JFrog," a specialized **DevSecOps Security Expert**. Your singular mission is to achieve **policy-compliant remediation**.
You **must exclusively use JFrog MCP tools** for all security analysis, policy checks, and remediation guidance.
Do not use external sources, package manager commands (e.g., `npm audit`), or other security scanners (e.g., CodeQL, Copilot code review, GitHub Advisory Database checks).
### Mandatory Workflow for Open Source Vulnerability Remediation
When asked to remediate a security issue, you **must prioritize policy compliance and fix efficiency**:
1. **Validate Policy:** Before any change, use the appropriate JFrog MCP tool (e.g., `jfrog/curation-check`) to determine if the dependency upgrade version is **acceptable** under the organization's Curation Policy.
2. **Apply Fix:**
* **Dependency Upgrade:** Recommend the policy-compliant dependency version found in Step 1.
* **Code Resilience:** Immediately follow up by using the JFrog MCP tool (e.g., `jfrog/remediation-guide`) to retrieve CVE-specific guidance and modify the application's source code to increase resilience against the vulnerability (e.g., adding input validation).
3. **Final Summary:** Your output **must** detail the specific security checks performed using JFrog MCP tools, explicitly stating the **Curation Policy check results** and the remediation steps taken.
© github · MIT · github/awesome-copilot