The Agent Skills Directory

[NO_CODE]: The skill consists solely of markdown documentation. It provides illustrative code snippets but no functional scripts, executables, or configuration files intended for the agent host environment.- [PROMPT_INJECTION]: The skill describes a methodology for processing untrusted conversation transcripts using an LLM, which identifies a surface for indirect prompt injection. If implemented as described, a malicious user in a conversation could attempt to manipulate the summary or influence the bot's behavior. Ingestion points: The skill describes processing 'conversation.full_transcript' within the 'ai_summarize' function in SKILL.md. Boundary markers: The example prompt templates do not utilize delimiters or specific instructions to ignore embedded commands within the transcript data. Capability inventory: The skill does not define or implement any tools, network operations, or subprocess executions. Sanitization: The documentation mentions privacy-related redaction but does not address sanitization against prompt injection attacks.- [SAFE]: No malicious instructions, obfuscated content, or unauthorized data access patterns were detected. The skill's behavior is consistent with its stated purpose as a developer guide.

conversation-summarization