The Agent Skills Directory

[SAFE]: The skill serves as a structural and linguistic analyzer for text drafts. Its operations are confined to reading specific state variables (book_state.chapter_laid_out, book_state.research_pack) and appending diagnostic reports to a internal log.
[SAFE]: No evidence of network exfiltration, remote script downloads, or unauthorized command execution was found. The instructions rely entirely on the agent's internal reasoning and local state variables.
[SAFE]: The skill implements a 'mirror not target' policy, explicitly instructing the agent to never modify or rewrite the source text. This reduces the risk of unintended data manipulation or integrity violations.
[SAFE]: No obfuscation techniques, such as Base64 encoding of commands, zero-width characters, or homoglyph attacks, were detected in any of the skill files.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it ingests untrusted text from the book_state object for analysis. However, since the skill has no access to dangerous tools (e.g., shell, network, or arbitrary file system writes) and produces only diagnostic JSON, the risk is negligible and consistent with its primary analytical purpose.

anti-slop