The Agent Skills Directory

[PROMPT_INJECTION]: The skill uses authoritative and directive language intended to override or prioritize specific behaviors, such as 'CRITICAL: THIS SKILL MANAGES YOUR MEMORY. WITHOUT IT, YOU WILL FORGET' and 'YOU MUST PROACTIVELY MANAGE YOUR HISTORY. Do not wait for the user to tell you.'
[INDIRECT_PROMPT_INJECTION]: The skill establishes an attack surface where the agent is encouraged to autonomously process and summarize potentially untrusted conversation history into new context states.
Ingestion points: Conversation history and tool outputs processed via context_log and context_checkout (SKILL.md).
Boundary markers: None specified in the instructions to distinguish between instructions and data during summarization.
Capability inventory: The skill utilizes context_tag and context_checkout to modify the agent's active memory and history (SKILL.md).
Sanitization: No explicit sanitization or validation steps are defined for the content being summarized.

context-management