The Agent Skills Directory

[DATA_EXFILTRATION]: The skill's primary function involves accessing and processing highly sensitive personal information.
It targets private communication logs from platforms such as WeChat, iMessage, Slack, Discord, Telegram, and Email (SKILL.md Phase 2).
It explicitly mentions the requirement for 'Full Disk Access' on macOS to access the iMessage SQLite database, which contains the user's entire message history (SKILL.md Phase 2).
[COMMAND_EXECUTION]: The skill documentation instructs users to execute external command-line tools to perform the data collection.
It provides specific commands for running an external Python script: python3 kit/immortal_cli.py collect --platform <平台> (recipes/steamer-flow.md).
[EXTERNAL_DOWNLOADS]: The instructions encourage users to download and install external code repositories from GitHub to enable full functionality.
It recommends using git clone for the agenmod/steamer-skill and agenmod/immortal-skill repositories (README.md).
[PROMPT_INJECTION]: The skill exhibits a significant surface for indirect prompt injection (Category 8) due to its processing of untrusted external data.
Ingestion points: The skill ingests arbitrary chat logs, emails, and exported data from various third-party platforms (SKILL.md Phase 2).
Boundary markers: There are no defined boundary markers or instructions to the AI to ignore malicious commands or directives that might be embedded in the ingested logs.
Capability inventory: The skill allows the agent to generate and write multiple new markdown files to the local file system based on the content of the ingested data (SKILL.md Phase 5).
Sanitization: There is no mechanism described for sanitizing, validating, or escaping the external content before it is processed by the LLM or written to disk.

steamer