The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/maintain_obsidian_course.py uses subprocess.run to execute Python scripts (collect_buaa_course_replays.py and extract_buaa_classroom.py) located in a neighboring skill directory (buaa-classroom-summarizer). The commands are passed as lists, which is a secure practice to prevent shell injection. This execution is part of the intended multi-skill workflow for processing course replays.
[DATA_PROCESSING_SURFACE]: The skill facilitates the reconstruction of course notes from untrusted external transcripts, presenting an indirect prompt injection surface.
Ingestion points: External transcripts and metadata are ingested in scripts/maintain_obsidian_course.py and scripts/review_final_note.py to generate note packets.
Boundary markers: The SKILL.md defines a strict 'Authoring Contract' and uses a generated semantic_rebuild_prompt.md to scope the agent's behavior during reconstruction.
Capability inventory: The skill has the capability to write Markdown files to the local Obsidian vault and execute local maintenance scripts.
Sanitization: The skill includes a dedicated validation script scripts/validate_final_note.py that checks for internal markers, generic boilerplate, and malformed timestamps, effectively gating low-quality or potentially poisoned outputs.
[EXTERNAL_DOWNLOADS]: The skill interacts with external course replay systems via secondary scripts to fetch academic content. This is the primary functional purpose of the tool and is handled through structured data exchange.

obsidian-course-vault