One registry for every AI agent skill your teams ship
SkillValit gives your workspace a single source of truth for agent skills: versioned, role-governed, and installable into Claude, Codex, or Gemini with one CLI command.
Free for small teams · No credit card required
Works with your stack
From messy prompt folder to governed registry in three steps
No migration project. Publish what you already have, put roles around it, and let any agent install it.
Publish
Add a skill with a name, description, markdown body, and tags. Every save auto-snapshots a version, so nothing you write is ever lost.
Govern
Assign viewer, editor, and owner roles. Every mutation is permission-checked and audit-logged. Group related skills into installable Skill Sets.
Install
One command drops a skill into any runtime you run. Use add-set to install a whole collection at once.
Everything a skill needs to be production-grade
Six capabilities, one registry. Built for teams that treat prompts like production code.
Versioned by default
Every save snapshots a full version. Roll back any skill instantly, with a complete history of who changed what.
One-command install
One command drops a skill into Claude, Codex, Gemini, or any agents.md runtime.
Skill sets
Bundle related skills into installable collections, so a new agent (or a new hire) is productive in minutes.
Role-based governance
Viewer, editor, and owner roles on every workspace. Every mutation is permission-checked and audit-logged.
Execution telemetry
See run counts, success rates, and latency per skill, so you invest in the skills that actually perform.
Dependency graph
Skills declare what they depend on. Visualize the graph and change with confidence, not guesswork.
You could keep your prompts in a git folder. Here is what that costs.
A shared repo of markdown files works until a second person, a second agent, or a bad edit shows up. SkillValit does the parts git was never built for.
Anyone with repo access can read and rewrite every prompt. Read-only or per-file permissions do not really exist.
Viewer, editor, and owner roles on every workspace. Every write is permission-checked before it lands.
Only as good as people's commit hygiene. Restoring one prompt to last week means digging through git log.
Every save auto-snapshots a version. Roll any skill back to any earlier version in one click.
Copy-paste into each agent's config by hand, or write and maintain a glue script per runtime.
One command installs a skill into Claude, Codex, Gemini, or any agents.md runtime.
A subfolder and a naming convention everyone has to remember and honor.
Named skill sets install as one collection with add-set. A new hire is productive in minutes.
Grep for references and hope you caught them all before you change a shared prompt.
Skills declare dependencies, rendered as a visual dependency graph you can change against with confidence.
A deploy key or PAT that usually grants far more than read-a-prompt, and rarely gets rotated.
Workspace-scoped API keys that only resolve skills, and can be rotated whenever you want.
Prompt edits are buried in the same git log as every other code change.
A dedicated audit log of who changed which skill, and when, separate from your codebase history.
Simple, flat pricing. One price per workspace
No per-seat math and no overage billing. Pick a plan for the whole workspace and upgrade inside the app whenever you outgrow it.
For small teams getting started with a shared skill registry.
500 installs / mo
- 1 workspace
- Up to 3 members
- 25 skills, 5 skill sets
- Version history, 5 deep
- CLI install to every runtime
For growing teams that ship and install skills every week.
10,000 installs / mo
- Up to 5 workspaces
- Up to 10 members
- Unlimited skills and skill sets
- Full version history
- Telemetry dashboard and dependency graph
For workspaces standardizing skills at scale, with SSO and export.
100,000 installs / mo
- Up to 10 workspaces
- Up to 50 members
- Everything in Team
- Domain SSO auto-join
- Audit export and higher rate limits
All plans start on Free. Billing is handled inside the app. No credit card to begin.
The questions a platform or security team asks first
Straight answers, including the ones where the honest answer today is "not yet."
Where does my skill content actually live?
Can I self-host SkillValit?
Are you SOC 2 certified?
Who can see and change a skill?
Does the CLI upload my local files when I install a skill?
agents.md). It reads from the registry down to your machine; it does not upload your local files or code back to SkillValit.What happens to my skills if I stop using SkillValit?
Still have a question we did not answer? Ask us directly.
Not ready to sign up?
We are launching on Product Hunt soon. Drop your email and we will send one note on launch day. No drip sequence, no spam.
One email, at launch. Unsubscribe in a click.
Put your agent skills under management
Bring your messiest prompt library. Publish it, govern it, and install it into any agent in an afternoon.