List an Agent

Any public GitHub repo that ships an ADL agent.yaml at its root can be listed in the catalog. Adding it is a one-line PR.

For listing a portable skill instead, see List a Skill.

Prerequisite: a public repo with agent.yaml at root

The catalog aggregator pulls agent.yaml directly from your repo. It must:

• Live at the root of the repo (not under examples/ or cmd/).
• Use apiVersion: adl.inference-gateway.com/v1 - the site rejects anything else.
• Carry valid metadata.name, metadata.description, and metadata.version (semver).

The full ADL schema, plus the adl CLI that scaffolds compliant repos, lives at adl.inference-gateway.com. This page does not re-document the schema.

Use the "+ Add agent" button

The fastest path is the + Add agent button on the Agents page. It opens a dialog with a copy-paste snippet and a "Continue to GitHub" link that lands you at https://github.com/inference-gateway/agents/edit/main/agents.yaml. GitHub will fork the repo and open the PR for you - paste the entry, commit, and submit.

The entry shape

agents.yaml is a flat list. Each entry is two lines:

- url: https://github.com/<owner>/<repo>
  ref: v1.0.0

• url is required and must be a https://github.com/... URL.
• ref is optional. Omitting it means latest: the build resolves your newest GitHub release tag, falling back to the newest git tag only if the repo has cut no releases at all. A repo with neither releases nor tags fails the build. Pinning an explicit release tag is recommended for third-party agents so a breaking change upstream cannot silently invalidate the catalog.

What happens after merge

The build-catalog.yml workflow in inference-gateway/agents runs on pushes to main that touch agents.yaml, scripts/build-catalog.mjs, package.json, package-lock.json, or the workflow file itself, plus workflow_dispatch. There is no cron. It:

1. Fetches agent.yaml from each repo at the listed ref (resolving latest).
2. Validates it against the ADL JSON Schema.
3. Rejects duplicate metadata.name collisions.
4. Sorts entries by name and writes catalog.json.
5. Opens or updates a chore(catalog): rebuild catalog.json pull request from the catalog/update branch - it does not auto-commit to main.

A maintainer merges that rebuild PR. Your entry then becomes visible here once inference-gateway/agents publishes a release, because this site reads the catalog at jsDelivr @latest (the repo's newest tag). Releases there are triggered manually.

Optional: make the card pop

The card pulls the following fields straight out of your agent.yaml. They are not required, but they make the listing more useful:

• spec.card.documentationUrl - turns on the Documentation button on the card.
• spec.deployment.cloudrun.image or spec.deployment.kubernetes.image - enables the OCI Image row so users can pin the exact image you publish. For repos outside the inference-gateway org this is the only way to surface an image at all.
• spec.tools[].tags and spec.skills[].tags - aggregate into the card's tag list and feed the tag filter on /agents/.
• spec.scm.url - explicit source URL; otherwise the aggregator's _source.url is used.