The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes several local utility scripts (scripts/bootstrap-vault.sh, scripts/ingest-url.sh, scripts/new-query-note.sh, and scripts/lint-wiki.py) to manage the file system and perform wiki maintenance tasks. These scripts are invoked via bash and are restricted to operations within the user-specified vault directory.
[EXTERNAL_DOWNLOADS]: The skill includes functionality to ingest external web content into the wiki using the scrapling utility. This is a primary feature of the skill, allowing the agent to capture raw source material from arbitrary URLs provided by the user.
[PROMPT_INJECTION]: The skill's workflow involving the summarization of external web pages and local files introduces a surface for Indirect Prompt Injection.
Ingestion points: External content enters the context via scripts/ingest-url.sh (URL scraping) and the raw/sources/ directory (local file ingest).
Boundary markers: The AGENTS.md file establishes a schema that instructs the agent to treat raw/ as an immutable source of truth and wiki/ as a maintained artifact, providing structural guidance to the model.
Capability inventory: The skill possesses capabilities for web fetching (WebFetch tool and Scrapling), file reading/writing, and executing local shell scripts.
Sanitization: There is no explicit sanitization of fetched content; the skill relies on the LLM's inherent processing and the structural rules defined in the schema to maintain safety.

llm-wiki