AI Agent
Skill Catalog

Create, version, approve, and publish AI Agent Skills in a shared catalog. Find reusable expertise with its owner, source evidence, and operating constraints attached.

Find the skill. Keep the context.

  1. 01 Find the expertise
  2. 02 Inspect the context
  3. 03 Reuse with ownership

01 / Keep the context

Give expertise a record teams can inspect

A team finding a skill needs to understand the work it supports, the reasoning it contains, and the conditions that limit its use. The source behind a rule and the person responsible for it belong in that same picture.

The AI Agent Skill Catalog organizes those details around a reusable company asset. Create and maintain the skill, inspect its evidence, manage review and approval, and publish a version that other teams can find.

A clear purpose

Connect the skill to the domain, decisions, and procedures it supports.

Traceable expertise

Keep reasoning, supporting sources, and operating constraints together.

Accountable reuse

Make ownership, version, and approval status part of the skill record.

02 / Find a skill

Find the skill that fits the work

Explore this illustrative catalog. Search by the work, narrow the domain, and select a skill to inspect the context that accompanies its published version.

The context travels with the skill
EvidenceOwnerSkillReuse
Source evidence and ownership stay connected to the skill as it becomes available for reuse.

3 example skills

Engineering / Illustrative skill recordPublished · v1.2

Release recovery

Support investigation of service errors following a deployment.

Owner
Service reliability lead
Decision logic
Compare the error timeline with the release and examine data compatibility before considering a recovery path.
Source evidence
Incident timeline, change record, and the application owner’s explanation.
Operating boundary
Request expert review when the timeline is incomplete or data compatibility is unresolved.
Service operations / Illustrative skill recordPublished · v2.1

Customer escalation

Identify when a customer issue needs specialist attention.

Owner
Customer support lead
Decision logic
Compare the issue with the documented escalation conditions and identify the team responsible for the next decision.
Source evidence
Escalation policy, resolved case examples, and the service lead’s explanation.
Operating boundary
Refer cases outside the documented conditions to the service lead.
Manufacturing / Illustrative skill recordPublished · v1.0

Quality inspection

Support examination of assembly deviations against defined inspection criteria.

Owner
Quality engineering lead
Decision logic
Compare the observed deviation with the documented criteria, material context, and examples before recommending a disposition.
Source evidence
Inspection criteria, annotated deviation examples, and the quality engineer’s explanation.
Operating boundary
Seek engineering review for unfamiliar materials or deviations outside the criteria.

03 / Govern the versions

Give every revision its own review

New evidence can change a decision rule, reveal an exception, or narrow a skill’s operating scope. Keep that change visible through versioning, review, and approval.

Describe what changed

Connect the proposed revision to the new finding or source. Show the reviewer which reasoning, procedure, or constraint needs examination.

Review the changed context

Examine how the update affects the skill’s intended use. The owner makes the approval decision for that version before publication.

Keep the published version identifiable

Make it clear which version is available for reuse and which is still being developed. Preserve the source context behind the version a team chooses.

Track review and readiness
Illustrative version history

A new draft starts a new review

v1.2PublishedReusev1.3DraftReviewApproval
The published version and a proposed revision have distinct statuses. A draft needs review and owner approval before it can be published.
Example change to examine

A new migration exception changes when the recovery skill should request expert review. The proposed version includes the explanation and supporting source for the owner to inspect.

Explore AI Skill Validation

04 / Prepare for reuse

Carry the context into the next workflow

A shared catalog gives teams a place to discover skills and examine their fit. Bring the selected version, its evidence, and its limits into the work where an AI agent will use it.

01 / Fit

Match the intended use

Compare the skill’s scope with the team’s workflow. Check the required inputs, operating conditions, and escalation boundaries before choosing it for reuse.

Explore skill engineering
02 / Exercise

Test the version in context

Use the AI Skill Playground to examine how the selected skill handles representative situations and exceptions. Keep its intended use and supporting evidence in view.

Explore the AI Skill Playground
03 / Improve

Bring new findings to the owner

When a source changes or an unfamiliar condition appears, return that evidence to skill development. Use it to inform the next version and its review.

Explore knowledge source intake

Start with expertise worth sharing

Which skill should another team be able to find?

Bring a workflow, the expertise behind it, and the teams that could reuse it. Explore how to give that capability a clear scope, supporting evidence, an owner, and a place in the catalog.

RapidKT Plain Text Website