The Agent Skills Directory

[SAFE]: The skill establishes strict confidentiality protocols, explicitly prohibiting the public disclosure of vulnerabilities or proof-of-concept code.
[SAFE]: Includes comprehensive data minimization instructions, mandating the redaction of credentials, tokens, and PII before any report generation or sharing.
[SAFE]: Recognizes the risk of untrusted data in security contexts, specifically instructing the agent to inspect submitted PoCs and execute them only in isolated environments when necessary.
[SAFE]: Mandates human approval for all external actions, ensuring that no reports are submitted or published without explicit user review of the content and recipient.
[SAFE]: References authoritative security resources and policies from trusted organizations like the OpenSSF, NIST, and GitHub to guide the discovery of safe reporting channels.

responsible-security-disclosure