The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted transcription files which could potentially contain instructions aimed at influencing the agent's behavior during the project naming and indexing stage.\n
Ingestion points: The skill utilizes list_directory and glob to scan user-specified directories and processes file contents via the doc_merger.py script.\n
Boundary markers: The doc_merger.py script mitigates injection risks by encapsulating ingested content within markdown code blocks.\n
Capability inventory: The skill has permissions to write files, execute shell commands for moving files, and run its own Python script.\n
Sanitization: Content is read directly from files with standard error replacement; no specific sanitization for malicious prompt sequences is performed on the data itself.\n- [COMMAND_EXECUTION]: The skill executes a bundled Python script and standard shell commands for file management and organization.\n
The doc_merger.py script is called to aggregate text files based on a configuration generated by the agent.\n
The mv shell command is used to rename output files according to extracted project titles.\n
These actions are focused on local file manipulation and align with the skill's stated purpose of document archival.

long-audio-to-obsidian