The Agent Skills Directory

[DYNAMIC_EXECUTION]: The script scripts/literature_cli.py implements a custom bootstrapping mechanism in the _bootstrap_ane_package function. This function dynamically modifies sys.path by searching for the ane_package directory across multiple locations, including environment variables, parent directories, and hardcoded user paths. This allows the script to load modules from paths computed at runtime.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted content from the OpenAlex API, such as scholarly titles and abstracts.
Ingestion points: External data enters through the OpenAlexClient and is formatted for display in scripts/literature_cli.py.
Boundary markers: Output lines are structured using | as a delimiter, and metadata like year and citation counts are explicitly separated from titles.
Capability inventory: The CLI tool's capabilities include printing to standard output and environment bootstrapping; the consuming agent manages the resulting data.
Sanitization: Titles are truncated to 90 characters in scripts/literature_cli.py, and the output streams are reconfigured to handle encoding errors using replacement characters.
[EXTERNAL_DOWNLOADS]: The skill is designed to interact with external scholarly databases.
Evidence: It performs network operations to the OpenAlex API and provides an optional escalation path to the Consensus search engine via the mcp__claude_ai_Consensus__search tool. Both are established services for academic literature discovery.

literature