The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted input from external files and user bio drafts. It incorporates defensive instructions to treat these as untrusted content and explicitly directs the agent to ignore any commands or instructions found within the data (Instruction 1). The indirect prompt injection surface is well-mitigated through these clear boundary instructions.\n- [DATA_EXFILTRATION]: All data operations are confined to the local project environment. The skill reads from memory/narrative-registry/ and writes to memory/narrative/ and memory/claims/. No network operations or external data transfer mechanisms are present.\n- [COMMAND_EXECUTION]: There is no evidence of shell command execution, subprocess spawning, or dynamic code generation. The skill's primary function is text processing and file writing within a restricted project scope.

narrative-enablement-kit