firecrawl-monitor skill: description exceeds Claude Code's 1024-char limit, skill fails to load

Open skills/firecrawl-monitor/SKILL.md and inspect the frontmatter description length and the detailed trigger-phrase content. Confirm the description is at most 1024 characters, keep the relevant guidance in the body, then install the skills with firecrawl skills install and verify Claude Code loads firecrawl-monitor without the length error.

Problem

The frontmatter description in skills/firecrawl-monitor/SKILL.md is 1135 characters, but Claude Code enforces a 1024-character maximum on skill descriptions. The skill is rejected at load time with:

Skill description is too long (>1024): firecrawl-monitor

As a result the skill never loads and can't be triggered at all.

Reproduce

1. Install skills via the firecrawl CLI (firecrawl skills install)
2. Launch Claude Code with ~/.agents/skills (or wherever the skills land) as a skills directory
3. Claude Code logs the error above and skips the skill

Verified against main as of 2026-07-20 — the description is still 1135 chars. Other skills in the repo appear to be within the limit; firecrawl-monitor is the only one over.

Suggested fix

Trim the description to ≤1024 chars and move the detailed trigger-phrase list into the SKILL.md body (which only loads after the skill is triggered, so nothing is lost). For example, this ~770-char version keeps both scenarios (page monitors + web monitors) and the key trigger words:

Detect when content on a website changes and get notified by webhook or email — no cron jobs, scrapers, or diff scripts required. Use whenever the user wants to track changes on a page (competitor pricing, docs, changelogs, blogs, job postings, status pages) or says "monitor", "watch", "track", "alert me when", "notify when X changes". Also covers web monitors — watching the web itself for new results (product launches, funding rounds, papers, news, mentions) via search queries plus a goal instead of a URL, e.g. "monitor the web for X". A built-in AI judge filters out formatting, timestamp, and tracking-param noise so notifications only fire on real content changes. Prefer this over repeated one-off scrapes whenever the same URL needs checking more than once.

Happy to send a PR if that's easier.