Three SKILL.md descriptions run past the documented 1,024 character limit

[kasparovabi]

Three skills in this repo carry a description longer than the 1,024 character maximum in the skill authoring docs, and eleven SKILL.md bodies run past the 500 line guideline. Reference: Agent Skills best practices.

Descriptions over 1,024 characters

File	description length
skills/bb-local-toolkit/SKILL.md	1,764
skills/bug-bounty/SKILL.md	1,405
skills/osint-methodology/SKILL.md	1,288

Each of these is a single-line description: value, so the count is the string itself with no folding involved. In bug-bounty and bb-local-toolkit the Chinese trigger words at the tail (中文触发词：漏洞赏金、安全测试...) sit inside the same budget.

I have not loaded these three skills into Claude Code to watch what the loader does past the limit, so I cannot tell you whether the value is truncated or the file is rejected outright. The character counts are what I can show.

Bodies over the 500 line guideline

File	body lines
skills/osint-methodology/SKILL.md	1,640
skills/bug-bounty/SKILL.md	1,599
skills/bb-local-toolkit/SKILL.md	1,553
skills/security-arsenal/SKILL.md	907
skills/web2-recon/SKILL.md	693
skills/supply-chain-attack-recon/SKILL.md	642
skills/web3-audit/SKILL.md	601
skills/report-writing/SKILL.md	563
skills/cloud-iam-deep/SKILL.md	515
skills/bb-methodology/SKILL.md	508
skills/hunt-ssrf/SKILL.md	508

Body lines are counted after the closing frontmatter marker. The four at the top of that table matter most, since the whole body loads into context the moment the skill triggers. The docs suggest moving the deep material into files under the skill directory and pointing at them from the body, so the payload tables and long command listings only get read when they are actually needed. bb-methodology, cloud-iam-deep and hunt-ssrf are a handful of lines over and probably not worth touching.

Two smaller things from the same pass. docs/verification/phase2h-smuggling-cachepoison.md is 177 lines with no table of contents, which the docs ask for above 100 lines so a partial read still shows the scope. And skills/hunt-sharepoint/SKILL.md:12 anchors its framing to "in 2025-2026 bug bounty / red-team work", which the docs suggest putting in an "old patterns" section rather than the main flow.

I opened the three long descriptions and spot-checked the line counts by hand before writing this, and I am happy to be wrong about any of it. What brought me here is a checker I maintain that reads published skill repos against the documented frontmatter limits, so I am saying that up front rather than leaving you to wonder. If you would rather not get this kind of report, tell me and I will not send another.