The Agent Skills Directory

[PROMPT_INJECTION]: The skill instructions do not contain any patterns attempting to override agent safety filters or bypass core instructions. Instead, it includes strong defensive rules (e.g., 'Pushback Rules') to prevent the agent from being coerced into providing a false positive fit.
[DATA_EXFILTRATION]: There are no commands or instructions that access sensitive local files (like SSH keys or AWS credentials) or exfiltrate data to external servers. The skill uses tools for search and database retrieval as intended for its primary function.
[OBFUSCATION]: The content is written in clear, plain-text markdown. No Base64, zero-width characters, homoglyphs, or other obfuscation techniques were found.
[REMOTE_CODE_EXECUTION]: The skill does not perform remote code execution. It references other internal skills (e.g., meanest-editor) for hand-offs, which is standard behavior for modular agent workflows.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied pitches, which represents an attack surface for indirect injection. However, the skill explicitly mandates 'Hard Gates' such as the 'Anti-Slop' check and the 'Anchor Article' verification, which require the agent to find external evidence to justify a verdict, significantly mitigating the risk of following instructions embedded within the pitch itself.
[DYNAMIC_CONTEXT_INJECTION]: The skill does not use the !command syntax or any other method to execute shell commands at load time.
[METADATA_POISONING]: The YAML frontmatter (name, description, when_to_use) is consistent with the instructions and purpose of the skill body.

journalist-fit-check