The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses several shell-based tools to manage the local wiki environment.
Executes git rev-parse to resolve the repository root.
Uses awk and grep to parse project configuration files and discover candidate wiki pages.
Invokes a local Python script (research_wiki.py) to perform logging and rebuild the query pack.
Dynamically interpolates user-provided arguments into a bash array for file targeting, which is a potential injection point if arguments are not strictly validated.
[EXTERNAL_DOWNLOADS]: Fetches paper information from well-known academic services.
Uses WebFetch to download summaries from alphaxiv.org (e.g., alphaxiv.org/overview/<arxiv_id>.md).
Uses curl to fetch XML metadata from the official arXiv API (export.arxiv.org).
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by ingesting and processing untrusted data from the web.
Ingestion points: External text is fetched from alphaxiv.org and arxiv.org (Step 2.3).
Boundary markers: Absent; there are no instructions to the agent to ignore potential malicious instructions embedded within the fetched paper summaries.
Capability inventory: The agent possesses Write and Edit tool permissions, which it uses to modify local markdown files based on the processed external content.
Sanitization: Absent; the skill relies on the LLM to summarize the fetched content directly into the wiki sections.

wiki-enrich