The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The verification script in SKILL.md fetches release artifacts, checksums, and cryptographic signatures from the official GitHub repository of the author (prompt-security/clawsec). This is part of a standard security practice for artifact verification.
[COMMAND_EXECUTION]: The skill provides a shell script for operators to verify the integrity of release artifacts. It utilizes common utilities such as curl for downloading, openssl for signature verification, and jq/shasum for integrity checking. These operations are scoped to a temporary directory and are used purely for security validation.
[DATA_EXFILTRATION]: While the specification mentions monitoring sensitive file paths (e.g., ~/.ssh, ~/.aws/credentials), these are explicitly listed as detection patterns for identifying exfiltration in external traffic, not for the skill to access them itself.
[PROMPT_INJECTION]: The skill instructions do not contain any patterns attempting to override agent behavior or bypass safety guidelines.
[SAFE]: The skill currently serves as a specification scaffold and does not include active runtime code or proxies that could perform unauthorized actions. It emphasizes safety requirements such as opt-in usage, secret redaction, and no automatic system mutations.

picoclaw-traffic-guardian