mirror of
https://github.com/mukul975/Anthropic-Cybersecurity-Skills.git
synced 2026-08-28 12:19:41 +03:00
Mahipal
9429fe9d5a
The README advertised "29 security domains" and its table listed 29 rows summing to 785 of 817 skills. Six domains were missing entirely and nine carried stale counts, because alias subdomains fold into their canonical form (security-operations into soc-operations, and so on). The table is now generated from the subdomain field in each skill's frontmatter: 34 domains, summing to 817. The six that were invisible are Threat Detection (7), Blockchain Security (2), Wireless Security (2), Privacy Compliance (2), Data Protection (1) and Purple Team (1). AGENTS.md - short, imperative instructions for agents working in this repository: the flat layout, which files load when, the validator commands, and the rule against reintroducing regex frontmatter parsing. Kept deliberately brief; long context files measurably degrade agent performance. SUPPORT.md - where to ask what, what actually moves fastest, and an honest statement that there is no SLA and some PRs have waited months.
47 lines
2.5 KiB
Markdown
47 lines
2.5 KiB
Markdown
# Support
|
|
|
|
This project is maintained by one person alongside other work. That shapes what support looks like here, so it is worth being direct about it rather than leaving you to guess.
|
|
|
|
## Where to go
|
|
|
|
| I want to… | Use |
|
|
|---|---|
|
|
| Ask how something works | [Discussions](https://github.com/mukul975/Anthropic-Cybersecurity-Skills/discussions) |
|
|
| Report a skill that is wrong, broken, or out of date | [Issues](https://github.com/mukul975/Anthropic-Cybersecurity-Skills/issues) |
|
|
| Propose a new skill | Issue first, then a PR — see [CONTRIBUTING.md](CONTRIBUTING.md) |
|
|
| Fix something yourself | Open a PR. This is the fastest path to change. |
|
|
| Report a security problem | See [SECURITY.md](SECURITY.md) |
|
|
|
|
Issues are for concrete defects: a command that does not work, a flag that does not exist, a framework ID that is wrong, a script that fails. Questions belong in Discussions, where other people can answer too and the answer stays findable.
|
|
|
|
## What to expect
|
|
|
|
There is no SLA. Response time varies with what else is happening; some pull requests have waited months, and that is a real cost I am working to reduce rather than a policy.
|
|
|
|
What moves fastest, in order:
|
|
|
|
1. A PR that fixes one thing and passes the validators.
|
|
2. An issue that names the skill, quotes the failing command, and shows the actual output.
|
|
3. Everything else.
|
|
|
|
What tends to stall: large multi-skill PRs, feature requests outside the skill-library format, and requests for bespoke integration help.
|
|
|
|
## Before you open an issue
|
|
|
|
```bash
|
|
pip install pyyaml
|
|
python tools/validate-skill.py skills/<skill-name>
|
|
```
|
|
|
|
If a skill fails validation, that output is the most useful thing you can paste. If you are reporting a procedure that does not work, say which version of the tool you ran and what it printed — "this does not work" is not actionable, and a wrong procedure in a security library is worth fixing properly.
|
|
|
|
## What this project is not
|
|
|
|
It is a reference library, not a product. It ships no service, holds no data, and makes no availability guarantee. Skills describe procedures for systems you own or are authorised to assess — see [SCOPE.md](SCOPE.md).
|
|
|
|
If you need something with a support contract behind it, this is not that, and I would rather say so plainly than have you discover it during an incident.
|
|
|
|
## Helping
|
|
|
|
The most useful contributions are unglamorous: correcting a procedure you actually ran, adding a framework mapping you verified, or telling me two skills are competing for the same request. All three make the library measurably better for everyone using it.
|