The Agent Skills Directory

[SAFE]: The skill facilitates loading task state from the .claude/handovers/ directory. This is a standard operational procedure for agent continuity and does not involve risky external communications, unauthorized privilege changes, or obfuscated code.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests content from handover files which could contain instructions that influence the agent's behavior. This is a known surface for indirect prompt injection if the files originate from an untrusted source.\n
Ingestion points: Local files within the .claude/handovers/ directory as specified in Step 1 and Step 2.\n
Boundary markers: The skill includes an explicit instruction to avoid reading the file body until the specific target is identified, reducing the risk of accidental context contamination.\n
Capability inventory: The skill utilizes Glob and Read operations to access and load handover documents.\n
Sanitization: No sanitization is performed as the content is meant to be interpreted as project context.

takeover