Discover1
Search the catalog by use case, platform, category, and tags. Start from trending skills or specific security-focused templates.
Academy Explained
How FerretForge Academy Works
Academy is the workflow for operationalizing AI skills - discover reusable configs, evaluate risk and quality, improve them, then publish and adopt the best variants for your platform.
Workflow at a Glance
Scan2
Each skill is analyzed for security findings and quality metrics before it is trusted in your workflow.
Improve3
Use structural hints and optional LLM suggestions to improve clarity, completeness, and platform alignment.
Publish4
Publishing requires a clean security baseline. Approved skills include score metadata and generated variants.
Adopt and Iterate5
Adopt the best-fit variant for your toolchain, track outcomes, and submit improvements as your needs evolve.
Key Features
Fork
Create your own copy of any published skill to customize. Forked skills maintain attribution to the original author.
Versioning
Published skills support versioning. Authors can publish updates while adopters can pin to specific versions.
Improvement Requests
Submit improvement suggestions for any published skill. If approved by the author, changes are merged with attribution.
Leaderboard
Skills and authors are ranked by composite score and adoption count. The leaderboard highlights the best skills per platform.
My Skills & Adoptions
Track skills you've published and adopted from a personal dashboard. Monitor adoption counts and improvement requests.
How Agents Talk to Academy
Agents connect through MCP, CLI, or direct API clients. Read-only procedures can discover skills, while write actions require an authenticated token tied to a user.
Agent Runtime
- MCP client inside IDE
- CLI automation in CI
- External app using HTTP
https
https
FerretForge API Layer
- academy.search
- academy.getBySlug
- academy.trending
- academy.adopt
- academy.publish
- academy.improve
- academy.requestSkill
routes
routes
Academy Core
- Skill registry + variants
- Security and quality scores
- Adoption and version history
Public Read Path
Use search/get/trending procedures to discover published and canonical skills without signing in.
Authenticated Write Path
Publish, adopt, improve, and version operations require a FerretForge account and an authenticated token.
Scoring Model
Each skill has a composite score built from three dimensions. This makes skill quality transparent before adoption and helps teams prioritize what to improve.
Security35%
Intelligence35%
Platform Fit30%
Composite = (Security x 0.35) + (Intelligence x 0.35) + (Platform Fit x 0.30)
What Happens Next
Start by browsing the public guide, then sign in to search skills, open details, and adopt variants into your own workspace.