The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill interacts with several well-known developer platform APIs to perform searches.
Interacts with GitHub's REST API at api.github.com for repository, code, and issue searches.
Interacts with the Hacker News Algolia API at hn.algolia.com for community discussions.
Interacts with the HuggingFace Hub API at huggingface.co/api for models and datasets.
Interacts with the Stack Exchange API at api.stackexchange.com for technical Q&A.
[CREDENTIALS_UNSAFE]: The skill follows secure credential management practices by avoiding hardcoded secrets.
Documentation in SKILL.md instructs users to manage API keys via .env files or environment variables.
Scripts like github_search.py and huggingface_search.py read credentials from environment variables (GITHUB_TOKEN, HF_TOKEN) or optional command-line arguments.
[INDIRECT_PROMPT_INJECTION]: The skill processes external search data using structured formats and basic sanitization.
Ingestion points: Data is ingested from the GitHub, Stack Overflow, Hacker News, and HuggingFace APIs across all search scripts.
Boundary markers: The skill returns data to the agent in a structured JSON format, providing clear separation between data and instructions.
Capability inventory: The skill is restricted to network-based data retrieval from trusted domains and does not possess capabilities for file modification or shell command execution.
Sanitization: The _strip_html utility in scripts/stackoverflow_search.py and scripts/hackernews_search.py removes HTML tags from external content snippets.

sn-search-code