skill0 ships agent-readable Markdown content — no executable code is bundled in this repository. The risk surface is therefore limited to:
- Skill content that the agent may follow verbatim and apply to user systems.
- Embedded commands or links (
x <mod>, URLs) inside SKILL.md files. - Generated artifacts (
skill0-cn-hk/) that mirrorskill0-cn/via ljh-sh/zhhz — these cannot differ from the source beyond character-set conversion, but should still be sanity-checked after regeneration.
Please do not file a public GitHub issue for security-sensitive reports.
Email: security@x-cmd.com (preferred). Use GPG if your report warrants it — the key fingerprint is published at https://x-cmd.com/security.asc.
A useful report includes:
- Path(s) of the affected file(s) under
skill0/(or the source module if cross-repo). - A reproducer: which agent + task triggers the bad behavior.
- The blast radius: does it require user input, a network call, or a specific shell?
- A suggested fix, if you have one.
| Acknowledgement | within 3 business days |
| Triage verdict | within 7 business days |
| Fix or mitigation | landed in a follow-up commit; coordinated disclosure on request |
| Credit | added to the release notes unless you ask to stay anonymous |
If the report is about a third-party tool referenced from a skill (e.g. x clawhub, x skill, agent-browser), we will route it to the upstream maintainer and copy you on the handoff.
- Bugs in upstream
x-cmdmodules: file at https://github.com/x-cmd/x-cmd or via the samesecurity@x-cmd.comalias — both reach the same maintainer. - Hallucinations or stylistic complaints about agent output: not security issues; use GitHub issues.
- README.md — repo philosophy.
- CONTRIBUTING.md — content conventions and regeneration workflow.
- skill0/core/x-cmd/SKILL.md — install options and the security-focused install guide.