The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted data in the form of meeting transcripts and audio files, which creates a surface for indirect prompt injection attacks.
Ingestion points: Meeting transcripts (text or file) and audio recordings provided by users (SKILL.md).
Boundary markers: No specific boundary markers or instructions to ignore embedded commands are present in the core behavior or rules.
Capability inventory: The skill has capabilities to execute shell commands (whisper CLI) and read/write files to a local directory (~/.openclaw/meetings/).
Sanitization: There is no evidence of input sanitization or validation to prevent malicious instructions within a transcript from being executed by the agent.
[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands to detect and run the Whisper CLI for audio transcription.
Evidence: Uses which whisper to check for the tool and whisper [file] --output_format txt to process audio files.
[EXTERNAL_DOWNLOADS]: The skill recommends the installation of external software from a well-known service if not already present on the system.
Evidence: Recommends running pip install openai-whisper for audio handling capabilities.

meeting-notes