Only the latest released version of skill-fetch receives security fixes. Please ensure you are running the most recent version before reporting a vulnerability.

Do not report security vulnerabilities through public GitHub Issues.

Please report security vulnerabilities using GitHub Security Advisories:

1. Go to https://github.com/girofu/skill-fetch
2. Navigate to Settings → Security → Advisories
3. Click New draft security advisory
4. Fill in the details: affected versions, description, reproduction steps, and impact
5. Submit the draft — only you and the maintainer can see it

• 48 hours: Acknowledgment of the report
• 7 days: Assessment and initial fix plan provided
• Coordinated disclosure: We will work with you on a disclosure timeline before publishing the advisory publicly

If you do not receive a response within 48 hours, please follow up by opening a non-sensitive GitHub Issue noting that you have submitted a security advisory and are awaiting a response.

skill-fetch is designed with the following security properties:

API keys and tokens are stored exclusively in the local file ~/.claude/skills/.fetch-config.json on the user's machine. This file is never read, transmitted, or uploaded by skill-fetch itself. Users are responsible for protecting this file with appropriate filesystem permissions (recommended: chmod 600).

When a skill is installed, skill-fetch records the SHA-256 hash of the installed files at installation time. This hash can be used to verify that the installed skill has not been tampered with after installation.

Before and after installation, skill-fetch performs a security scan across 6 categories (labeled A through F):

• A: Credential and secret exposure (API keys, tokens, passwords in code)
• B: Dangerous shell operations (eval, arbitrary code execution, unsafe redirects)
• C: Network exfiltration (unexpected outbound connections, data exfiltration patterns)
• D: Filesystem access (writes outside expected skill directories, path traversal)
• E: Supply chain risks (unpinned dependencies, obfuscated code)
• F: Privilege escalation (sudo usage, setuid, capability manipulation)

Scan results are displayed to the user before installation is finalized, allowing an informed decision.

All shell scripts in skill-fetch use set -euo pipefail at the top of every script. This ensures:

• -e: Exit immediately on any command failure
• -u: Treat unset variables as errors (prevents silent use of empty strings)
• -o pipefail: Propagate errors through pipes rather than silently swallowing them

All API scripts use node's built-in JSON.stringify() and JSON.parse() for constructing and parsing JSON data. User-supplied query strings are never concatenated directly into JSON payloads or shell commands, preventing injection attacks.

This security policy covers vulnerabilities in skill-fetch itself — the plugin code, search logic, installation scripts, and security scanning infrastructure.

It does not cover:

• Security issues within skills that are discovered and installed by skill-fetch (those should be reported to the respective skill authors)
• Vulnerabilities in external registries or APIs that skill-fetch queries
• Issues arising from user misconfiguration of API keys or file permissions