The Agent Skills Directory

[SAFE]: The skill performs legitimate knowledge management tasks by analyzing internal project documentation and metadata files. No malicious intent, obfuscation, or dangerous execution patterns were identified.- [PROMPT_INJECTION]: Evaluated for indirect prompt injection potential due to data ingestion. * Ingestion points: Reads potentially untrusted data from workflow traces at docs/harvest/<workflow_id>.md and issues lists at docs/issues-filed.json (SKILL.md). * Boundary markers: Absent; the skill does not explicitly wrap input data in safety delimiters. * Capability inventory: Limited to substrate-read and substrate-write tools, which are restricted to internal documentation management (SKILL.md). * Sanitization: No explicit sanitization or escaping of input data is noted prior to processing, though the resulting actions are low-risk documentation writes.

harvest