VigilAgent — Security skill for Claude Code
Security auditing CLI for AI-agent-authored code changesets: detects package hallucinations, indirect prompt injection, and AI-specific code defects before they hit production.
How to install VigilAgent
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open sunilgentyala/VigilAgent and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What VigilAgent does
Security auditing CLI for AI-agent-authored code changesets: detects package hallucinations, indirect prompt injection, and AI-specific code defects before they hit production.
Alternatives in Security
- Security Guardian — Comprehensive security analysis for RTK CLI tool, focusing on command injection, shell escaping, hook security 11.9k ★
- Promptmap — Security scanner for custom LLM apps 1.2k ★
- Agentseal — Security toolkit for AI agents 344 ★
README
VigilAgent
[](https://github.com/sunilgentyala/VigilAgent/actions/workflows/ci.yml) [](LICENSE) [](package.json) [](https://sunilgentyala.github.io/VigilAgent/) [](CHANGELOG.md)
Security auditing CLI for AI-agent-authored code changesets: diffs produced by Claude Code, Cursor, Copilot Workspace, and similar coding agents.
VigilAgent reads a unified diff and runs three security-focused detectors that none of the existing terminal diff/review tools cover, then renders the findings as colored terminal output, JSON, or Markdown, with an optional CI exit-code gate.
Why VigilAgent
Terminal diff and review tools have converged on visualization and review ergonomics, not security. Agent-oriented diff viewers make it easier to follow what an agent changed, but they don't audit content. Structural and syntax-highlighting tools render diffs beautifully but perform no security checks at all. "AI slop" scanners catch style and dead-code smells, such as narrative comments and oversized functions, not security defects.
VigilAgent is the layer none of them are: a detector built specifically for the failure modes that show up when an agent, not a human, authored the change. It closes four gaps that are specific to AI-agent-authored code:
- Package hallucination: coding agents occasionally invent
plausible-sounding package names that don't exist on the real registry,
or get talked into adding a typosquatted lookalike. Nothing cross-checks
newly added
package.json/requirements.txt/Cargo.tomlentries against the live npm/PyPI/crates.io registries. - Indirect prompt injection in generated code: an agent
Related Skills
Kiteguard
Open-source runtime security guardrails for Claude Code, Cursor, and Gemini CLI — blocks dangerous commands, d
Patronus Security CLI
Security scanner and runtime protection for AI coding agents. Detects prompt injection, secrets, PII, and unsa
Skillguard
Security scanner for AI agent skills. Detects prompt injection, data exfiltration, and malicious payloads befo
LLM Security
AI/LLM agent security audit (OWASP GenAI/LLM Top 10 2026): prompt injection (direct + indirect/RAG), jailbreak
Agent Security Super Skill
Comprehensive AI agent security skill — prompt injection defense, skill validation, memory poisoning preventio
Inkog
Static security scanner for AI agents. Catches prompt injection, runaway loops, missing oversight, and complia
Related Agents
LLM AI Hunter
LLM and Agentic AI vulnerability specialist. Covers OWASP LLM Top 10 v2025 (LLM01-LLM10) and OWASP Agentic AI
Ehs AI Safety
AI, agent and chatbot security specialist for the Ethical Hacker Squad. Reviews the instruction/data boundary,
Kavach LLM
KAVACH LLM/AI security specialist. Audits prompt injection (direct + indirect/RAG), jailbreak/persona/system-p
