The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains phrases typically associated with prompt injection, such as 'ignore previous instructions', within its 'Prompt Injection Safety Notice' (Section 8). However, these are used as explicit defensive instructions to the agent to disregard malicious directives found in user-supplied logs or queries. This is a best-practice safety implementation rather than an attack vector.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data, including log samples and query drafts. It provides mandatory safety constraints to treat this content as data rather than commands and includes instructions to use placeholders to prevent leaking sensitive production data.
[SAFE]: The skill does not attempt network connections, file system modifications beyond reading logs, or credential harvesting. The YAML frontmatter correctly scopes the skill using 'allowed-tools' (Read, Grep, Glob), and the overall content is consistent with its stated purpose of security engineering support.

siem-rules