The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes multiple shell-based operations, including python modules (e.g., ane_package.orchestration.programmes, ane_package.feedback), git for repository synchronization, and npx for managing skill installations.\n- [EXTERNAL_DOWNLOADS]: During the consolidation process, the skill runs npx -y skills add gasserane/personal-skills. This command retrieves and installs executable skills from a remote repository at runtime.\n- [DATA_EXFILTRATION]: The skill is configured to perform git push on a repository containing synthesized knowledge artifacts and system improvements, providing a mechanism for local data to be uploaded to an external platform.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from documents (PDF, DOCX, HTM) and Microsoft Forms, creating a potential surface for instructions embedded in those documents to influence the agent.\n
Ingestion points: Document library on OneDrive, local wiki raw/ folders, and external MS Forms submissions.\n
Boundary markers: None provided; ingested content is not wrapped in protective delimiters.\n
Capability inventory: Subprocess execution (Python, Git, NPX), file system modifications, and network-based DOI verification.\n
Sanitization: Lacks general filtering or sanitization of ingested text to remove potentially malicious instructions.\n- [DYNAMIC_EXECUTION]: Uses python -c to execute code strings and interacts with a local Python package (ane_package) for core functionality, including orchestration and feedback processing.

li