The Agent Skills Directory

[DATA_EXFILTRATION]: The skill is designed to access and extract data from sensitive local locations, such as macOS iMessage databases and WeChat SQLite files. This access is a core requirement for its stated purpose of persona distillation from chat history.
[COMMAND_EXECUTION]: Utilizes local Python and shell scripts to manage data processing, metadata initialization, and version control. Key scripts include preprocess.py for database extraction and version_manager.py for state snapshots.
[EXTERNAL_DOWNLOADS]: References and executes external tools via npx and specialized CLI utilities for skill pack generation and publishing. These tools are associated with the vendor's ecosystem for AI persona management.
[PROMPT_INJECTION]: Acts as a surface for indirect prompt injection by ingesting and processing untrusted data from web search results and external file exports during persona extraction.
Ingestion points: Phase 3 (WebSearch results, chat log exports, PDFs, and social media archives).
Boundary markers: No explicit boundary markers or instructions to ignore embedded instructions are present in the processing logic.
Capability inventory: Includes Bash, Write, Edit, Read, and WebSearch tools.
Sanitization: Implements basic PII redaction for phone numbers and addresses but does not sanitize content for adversarial instructions.

anyone-skill