The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill includes Python scripts designed to read and process meeting transcripts exported from third-party services like Otter.ai and Fireflies.ai. This represents a potential ingestion surface for untrusted data that could contain malicious instructions.
Ingestion points: SKILL.md (Step 4 archival script) and references/implementation.md (Conversion scripts) read files from user-defined directories.
Boundary markers: Not explicitly present in the provided scripts; data is interpolated directly into markdown templates.
Capability inventory: The skill uses Bash to execute Python scripts which perform file system operations (Read, Write, Edit).
Sanitization: The scripts perform basic string interpolation of file content into markdown files. The risk is mitigated by the fact that the output is intended for manual archival and does not automatically feed into other high-privilege tool chains.
[SAFE]: The skill uses safe patterns for secret management by documenting the use of environment variables for API authentication in examples.
[SAFE]: The Python scripts use standard libraries to perform local file management tasks (renaming, moving, and formatting text) consistent with the skill's stated purpose of data migration.

granola-migration-deep-dive