skill-manager

Install, audit, update, and drift-detect agent skills from remote Git repositories.

Updated Dec 20, 2025
One-click install
npx skills add https://github.com/cristoslc/LLM-personal-agent-patterns --skill skill-manager-cristoslc
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: skill-manager
Source: https://github.com/cristoslc/LLM-personal-agent-patterns/tree/main/L3-agents-core/.agents/skills/skill-manager
Command: npx skills add https://github.com/cristoslc/LLM-personal-agent-patterns --skill skill-manager-cristoslc

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve? Managing agent skills across projects is error-prone: skills fetched from remote repos lose provenance, silently drift from upstream, and may contain unsafe shell patterns. This Skill provides full-lifecycle management with provenance tracking, security auditing, and drift detection. ## Core Features & Use Cases - Safety-gated installation: Installs skills via npx skills or a POSIX fallback, stamps a .source.yml provenance manifest, runs a security audit, and rolls back automatically on critical findings. - Security auditing: Scans skill directories for exfiltration, credential access, reverse shells, curl-pipe-shell, obfuscation, and prompt injection patterns. - Drift detection and updates: Compares SHA-256 integrity digests against recorded provenance to detect local modifications or upstream changes, and re-installs from recorded coordinates to update. - Use Case: You want to install a community code-review skill into your project. The Skill interviews you for scope and version, installs it, audits it for malicious patterns, stamps provenance, and later tells you when the upstream repo has changed. ## Quick Start Install the code-review skill from a GitHub repo into my project's .agents/skills directory and audit it for security issues.

Frequently Asked Questions about skill-manager

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I install an agent skill from a GitHub repository?▼

Run scripts/install.sh with the repo URL, skill path, optional ref, and target directory. It tries npx skills add first, falls back to a POSIX git-clone path, stamps a .source.yml manifest, and audits the result before activating.

How do I check if an installed skill was modified locally or changed upstream?▼

Run scripts/drift.sh on the skill directory to compare its current SHA-256 content digest against the digest recorded in .source.yml. Use --all to scan every skill in a directory or --cross to compare two projects.

What security risks does the skill audit detect?▼

The audit scans for data exfiltration via curl or wget, environment and credential harvesting, base64 obfuscation, eval usage, reverse shells, curl-pipe-shell patterns, prompt injection in Markdown, and known malicious commands like rm -rf /.

Does skill installation work without Node.js or npx?▼

Yes. When npx is unavailable or fails, install.sh falls back to fetch-remote-skill.sh, which uses only git, tar, sha256sum or shasum, and standard POSIX tools to clone, extract, and stamp the skill.

How do I update an installed skill to a newer version?▼

Run scripts/update.sh on the skill directory. It reads the repository, ref, and path from .source.yml, re-runs the safety-gated install, and reports whether the integrity digest changed or the skill is already up to date.

What happens when the audit finds critical security issues during install?▼

The installation is rolled back automatically: the new skill directory is removed and any previous version is restored from backup. The script exits with code 2 so callers can detect the rollback.