Agent Skills — /.well-known

The canonical agent surface for Harbor. Ten markdown files that teach an LLM how to use Harbor end-to-end — what to install, how to authenticate, which tool to call when, when to publish a Function, when to fall back to hrbr exec.

These are the source of truth for agent behavior. Every Harbor docs page is downstream of these files.

The skill set

(All paths above are relative to https://docs.tryharbor.ai.)

Discovery

Agent hosts that follow the Agent Skills convention discover Harbor skills automatically when the user mentions Harbor or tryharbor.ai. The host fetches SKILL.md first, then loads the topic skills referenced inside it on demand.

For hosts without auto-discovery (most current LLM clients), point them at the URLs in your system prompt or skill loader.

Frontmatter convention

Each skill starts with YAML frontmatter:

---
name: harbor-plugins
user-invocable: false
description: |
  Provides the agent's safe path to the workspace's connected
  SaaS / API / MCP / database tools through Harbor.
---

• name — the slug (= the filename without .md)
• user-invocable — false for topic skills; the user doesn't say "use harbor-plugins skill", they describe their intent
• description — used by the host's skill router

Cache headers

content-type: text/markdown; charset=utf-8
cache-control: public, max-age=300, s-maxage=86400, stale-while-revalidate=2592000

Same as /llms.txt — aggressive cache, low rebuild rate.