The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses npx to execute its own helper scripts and CLI. This is the primary mechanism for distributing and running the skill's logic.
[EXTERNAL_DOWNLOADS]: The skill instructions and documentation reference official GitHub repositories (github.com/nomadamas/*) and official RISS domains for documentation and API access. These are documented neutrally as they represent the vendor's official infrastructure.
[DATA_EXPOSURE]: The Python helper script (keris_academic.py) reads a local configuration file at ~/.config/k-skill/secrets.env to retrieve the user's RISS API key. This is a standard and safe practice for secret management, avoiding hardcoded credentials.
[CREDENTIALS_SAFE]: The skill explicitly instructs users NOT to print or store plaintext credentials in chat or shell arguments. It includes a --dry-run mode that redacts API keys in output to prevent accidental exposure.
[INDIRECT_PROMPT_INJECTION]: The skill ingests academic metadata from the RISS XML API.
Ingestion points: XML parsing of search results in scripts/keris_academic.py.
Boundary markers: The output is formatted into structured text or JSON, providing a clear boundary between search results and agent instructions.
Capability inventory: The skill is limited to searching and displaying metadata; it lacks capabilities for writing files, performing arbitrary network requests, or executing subprocesses based on search results.
Sanitization: The script performs validation on input fields (length and character checks) and uses standard XML parsing libraries.

keris-academic-search