agentsec-ai 0.4.1

Security scanner and hardener for agentic AI installations — OpenClaw, MCP servers, and AI agent skill ecosystems

agentsec

Security scanner and hardener for agentic AI installations.

agentsec focuses on practical misconfigurations and supply-chain risk in:

• OpenClaw installations
• MCP server configurations and code
• Agent skill/plugin ecosystems
• Credential exposure in local files

All findings map to the OWASP Top 10 for Agentic Applications (2026).

Quick Start

pip install agentsec-ai
agentsec scan

Scan a specific installation:

agentsec scan ~/.openclaw

What agentsec checks

Module	Scope
installation	Gateway exposure, auth posture, DM/group policy, tool/sandbox settings, SSRF and safety checks, known CVE version checks, sensitive file/dir permissions
skill	Instruction malware, risky code patterns (eval/exec/subprocess), prompt-injection patterns, frontmatter capability risk, dependency/install-hook risk
mcp	Tool poisoning patterns, auth gaps on remote endpoints, dangerous schema/permissions, unverified npx usage
credential	16 provider-specific secret patterns, high-entropy detection, git credential leakage

Reference catalog:

• checks-catalog.md (27 named checks + dynamic credential findings)

Core Commands

# Full scan
agentsec scan

# JSON output for CI
agentsec scan -o json -f report.json

# SARIF for GitHub Code Scanning
agentsec scan -o sarif -f results.sarif

# Run only selected scanners
agentsec scan -s installation,mcp

# Control CI failure threshold
agentsec scan --fail-on critical

# Hardening preview (default: preview only)
agentsec harden -p workstation

# Apply hardening profile
agentsec harden -p workstation --apply

# Watch mode
agentsec watch ~/.openclaw -i 2

# Pre-install security gate
agentsec gate npm install some-skill

List available scanners:

agentsec list-scanners

Hardening Profiles

Profile	Intended use
workstation	Single-owner local usage
vps	Remote/self-hosted deployment
public-bot	Highest restriction for untrusted public input

Use agentsec show-profile <name> to inspect exact changes before applying.

Output and Exit Codes

Output formats:

• terminal (default)
• json
• sarif

Exit codes:

• 0: no findings at/above threshold
• 1-127: count of findings at/above threshold (capped)
• 2: runtime/usage error

GitHub Actions

Use the bundled composite action:

name: Agent Security
on: [push, pull_request]

jobs:
  agentsec:
    runs-on: ubuntu-latest
    permissions:
      contents: read
      security-events: write
    steps:
      - uses: actions/checkout@v4
      - uses: debu-sinha/agentsec@v0.4.1
        with:
          fail-on: high
          output: sarif
          upload-sarif: 'true'

Action definition:

• action.yml

Screenshots

Step 1: Scan an insecure installation

Step 2: Apply hardening profile

Step 3: Re-scan after hardening (14 findings remain — credentials need manual rotation)

Benchmarks and Studies

• Fixture benchmark (v0.4.0, dated 2026-02-15)
  Precision/recall/F1 over a 20-fixture suite.
• Top-50 MCP study (snapshot 2026-02-16)
  Agentsec-only repro run with normalized findings output.
• Top-50 study kit
  Schema, selection CSV, JSONL findings, and summary JSON.

Current checked-in Top-50 summary data:

• docs/benchmarks/top50/reports/top50_summary_20260215.json
• docs/benchmarks/top50/reports/top50_findings_20260215.jsonl

Reproducibility

Fixture benchmark:

python docs/benchmarks/run_benchmark.py

Top-50 study repro:

powershell -ExecutionPolicy Bypass -File scripts\reproduce_top50_study.ps1 -DateStamp 20260215

Note: semgrep and gitleaks baseline runs are optional in the PowerShell repro script and require those tools on PATH.

Adversarial consistency audit (docs vs artifacts):

python scripts/repo_consistency_audit.py

Case Studies

• 001: Insecure workstation remediation
• 002: Public bot hardening on VPS
• 003: Pre-install gate blocked malicious package
• 004: Malicious skill detection and block

Launch Evidence

• Launch Evidence Manifest
• Reproducibility Spec

Development

git clone https://github.com/debu-sinha/agentsec.git
cd agentsec
pip install -e ".[dev]"
ruff check src tests
pytest

Contribution guide:

• CONTRIBUTING.md

Security policy:

• SECURITY.md

License

Apache-2.0