The Agent Skills Directory

[SAFE]: The skill operates entirely through structured prompts and predefined markdown schemas to perform literary analysis. A comprehensive review of the 31 provided files revealed no malicious patterns such as obfuscation, credential exposure, persistence mechanisms, or unauthorized command execution. The skill is purely documentation-based and does not include any executable scripts or tool-use definitions.
[SAFE]: An analysis for indirect prompt injection was performed because the skill audits untrusted drafts. 1. Ingestion points: chapter drafts and outlines via the context_bundle field. 2. Boundary markers: No specific delimiters are defined in the instructions, though the use of YAML schemas provides inherent structure. 3. Capability inventory: The skill has no access to shell commands, local file systems, or network APIs. 4. Sanitization: No sanitization of the audited text is mentioned. The risk is assessed as safe because the lack of system-level capabilities prevents any potential injection from causing harm.

novel-continuity-auditor