opencli-sitemap-author
Use when creating or maintaining OpenCLI site sitemaps: agent-facing navigation, page-state, action, workflow, API-reference, pitfall, and fallback knowledge for a website. Use after browser exploration discovers durable site context, when a sitemap is stale, or when promoting local site knowledge into the repo.
Security Assessment
About opencli-sitemap-author
opencli-sitemap-author is a skill for creating and maintaining OpenCLI "sitemaps" — but not SEO sitemaps. The artifact it produces is a task execution graph for agents that drive a website through `opencli browser`: it captures where an agent is on a site, which navigation paths to take, which OpenCLI adapter to prefer, and how to recover when the live page disagrees with stored memory. It solves the problem of durably recording verified, task-relevant site knowledge so future agents don't have to rediscover a site's structure from scratch.
The skill defines a two-layer storage model (a global seed under `sitemaps/<site>/` and a local overlay under `~/.opencli/sites/<site>/sitemap/`, where local wins by stable id) and a recommended file layout for site overview, per-page state signatures, cross-page UI partials, workflows, and pitfalls. It prescribes an authoring loop: load existing memory, verify against live browser state (browser state is treated as ground truth), record only durable structure using stable ids, write a local draft, and mark items stale on conflict. A required action schema captures preconditions, the action, success proof, failure signals, and recovery instructions including an adapter health writeback. A companion reference file specifies field-level constraints, an 800-token-per-file size budget, visual-anchor types, and rules against fragile selectors. Its declared tools are limited to `opencli` CLI calls plus Read, Edit, Write, and Grep.
Target users are agent authors and power users building reusable browser-automation knowledge for OpenCLI, particularly after a browser-exploration session discovers stable site context or when an existing sitemap has gone stale. Note that portions of the documentation are written in mixed English and Chinese.
FAQ
Is this an SEO sitemap generator?
No. It authors a task execution graph for agents using `opencli browser` — page states, actions, workflows, adapter preferences, and recovery paths — not an SEO or crawler sitemap.
Where are sitemaps stored?
In two layers: a global seed at `sitemaps/<site>/` and a local overlay at `~/.opencli/sites/<site>/sitemap/`. New discoveries are written to the local overlay first; the local layer wins by stable id, and content is promoted to global only after review.
What tools does the skill use?
It declares `Bash(opencli:*)`, Read, Edit, Write, and Grep — it runs OpenCLI browser commands to verify page state and edits local Markdown sitemap files.
What is the file size guidance?
The schema targets 800 tokens per file body (excluding frontmatter). Pages with many actions or workflows should be split into sub-files, and CJK content is counted by tokens rather than bytes since it runs 30-50% higher tokens-per-character.
How does it handle stale information?
Live browser state is treated as truth. When an existing sitemap disagrees with current browser state, the author trusts the browser and marks the conflicting item stale rather than forcing the old path.
Install opencli-sitemap-author
Quick Setup:
- Copy the skill folder to
.claude/skills/ - Claude will automatically detect and use the skill
Repository
jackwener/opencli