Reusable skills for your AI agents, scanned before they ship.
Skills package instructions, workflows and capabilities that AI agents reuse across tasks. Bud SENTRY scans every skill before it can be published, and you decide which projects and agents can use it.
From a skill bundle to an agent, in one walkthrough.
Import a skill bundle from GitHub, review the SENTRY security report, choose where it is deployed, and attach it to an agent in the Agent Builder.
Write a capability once. Every agent can use it.
Without a shared library, each team rewrites the same instructions into each agent, and nobody can say which version is running where. A skill is packaged once, checked once, and attached wherever it is needed, so agents work the same way across the enterprise and new agents start from capabilities that are already approved.
Five steps from bundle to agent.
Every skill follows the same path in the Skills tab, so nothing reaches an agent without being validated, scanned and given a scope.
-
01
Register
Click Register Skill and upload a ZIP bundle, import from a GitHub repository or paste a direct URL. Private repositories take an access token. Tags keep skills easy to find.
-
02
Review
Bud fetches the bundle, discovers its skills, validates each one and stages the bundle. Import every skill it contains, or select only the ones you need.
-
03
Scan
Bud SENTRY checks the selected skills for security threats, using static analysis or an AI-based analyser, and returns a security report.
-
04
Publish
Choose the scope, one project or the whole platform, and publish. The skill's instructions, licence and scan report stay open for review.
-
05
Attach
In Projects, open an agent in the Agent Builder and attach the skills it needs. The agent can then use them as part of its workflow.
Every file is checked before a skill is published.
A skill can carry code and instructions an agent will follow, so Bud SENTRY inspects each one before it can reach an agent.
Malware signatures
Every file in the skill is checked against ClamAV signatures.
Sandboxed execution
The skill is run inside a sandbox, so its behaviour is observed before it touches your environment.
Prompt injection
Instructions are analysed for attempts to override what the agent has been told to do.
Data exfiltration
The scan looks for paths that would send data out of your environment.
Privilege escalation
The scan flags code that tries to gain access beyond what the skill should have.
Credential exposure
The scan checks for secrets and credentials the skill could leak.
A report you can read before you decide.
When the scan finishes, Bud shows an overall assessment with the reasons behind each finding, so a reviewer can see why a skill was rated the way it was and decide whether to deploy it. In the demo, the selected skill is rated Low Caution, with an explanation for each finding.
You decide who can use each skill.
After reviewing the report, choose where the skill is deployed before you publish it.
One project
Only the agents in that project can use the skill. Use it for capabilities that belong to one team or one use case.
The whole platform
The skill is available more broadly across Bud AI Foundry, subject to the platform's access controls. Use it for capabilities every team should share.
After publishing, open any skill to see its Markdown instructions file, its licence and its security scan report, so you always know what a skill does and how it was assessed.
Where Skills fits.
Skills is part of Bud AI Foundry, and its security scan is Bud SENTRY's work.
Product
Bud AI Foundry
The control panel Skills lives in
Inference, guardrails, observability, evaluations and cost control on one serving plane, with agents built and governed in the same place.
Explore Bud AI FoundryProduct
Bud SENTRY
The security layer behind every scan
Zero-trust security, governance and compliance for the whole platform: every model verified at ingestion, every call classified both ways, one immutable audit trace.
Explore Bud SENTRYSee Skills on your own agents.
Bring a skill bundle you already use. We will import it, show you the security report and attach it to an agent on your infrastructure.
