The Agent Skills Directory

[SAFE]: The skill is a documentation utility that processes text from the conversation history. It does not contain instructions for privilege escalation, network exfiltration, or persistence. All file operations are directed to the docs/ directory for non-executable markdown storage.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from user conversations to build a glossary, which is a common surface for indirect prompt injection. 1. Ingestion points: Current conversation and the docs/UBIQUITOUS_LANGUAGE.md file. 2. Boundary markers: None. 3. Capability inventory: Limited to reading and writing a local markdown file; no shell execution or network tools are permitted. 4. Sanitization: No explicit content filtering or validation is performed on the extracted domain terms.

snap-lang