Skip to content

Security: AlterLab-IEU/AlterLab-Academic-Skills

Security

SECURITY.md

Security Policy

Reporting a Vulnerability

If you discover a security vulnerability in AlterLab Academic Skills, please report it responsibly.

Report via: Open a GitHub issue on this repository (use the "Security" label for sensitive reports).

Please include:

  • Description of the vulnerability
  • Steps to reproduce
  • Affected skill(s) and file path(s)
  • Potential impact assessment

Response Timeline

  • Acknowledgment: Within 48 hours
  • Assessment: Within 5 business days
  • Fix/Mitigation: Based on severity

Scope

Security concerns for this project include:

  • Prompt injection: Skills that could be manipulated to execute unintended actions
  • Data exfiltration: Skills that could leak sensitive information
  • Path traversal: Skills with hardcoded or manipulable file paths
  • Credential exposure: Skills that might expose API keys or tokens

Skill Security Scanning

All pull requests that modify skills are scanned for potential security issues. We recommend:

  • Never hardcode file paths, API keys, or credentials in skills
  • Use environment variables for sensitive configuration
  • Validate all external input referenced in skill instructions
  • Follow the principle of least privilege for allowed tools metadata

Trust Manifest

SECURITY_SCAN.md is a generated, CI-attested manifest of the shipped skill code. It enumerates the full outbound-host allowlist (all legitimate scientific/academic APIs, user-keyed LLM backends, or documentation placeholders — no telemetry) and attests, from a static scan, that the code contains no shell-pipe, no os.system / shell=True, no eval/exec on input, and no hardcoded secrets.

It is regenerated with python3 scripts/gen_security_scan.py and enforced in CI: --check fails the build if the manifest drifts from the code, and --strict fails if any attestation is violated. So the manifest can never silently go stale or become untrue — it is a verifiable signal you can check before installing, not a hand-maintained promise.

Supported Versions

Version Supported
v1.0.0 (Latest) Yes

There aren't any published security advisories