The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill retrieves literature metadata from well-known academic services including OpenAlex, Crossref, and Semantic Scholar.
Evidence: API requests are implemented in scripts/providers/openalex.py, scripts/providers/crossref.py, and scripts/providers/semantic_scholar.py using the requests library.
[COMMAND_EXECUTION]: The skill provides utility scripts for running searches and validating results locally.
Evidence: scripts/search_runner.py acts as a CLI runner, and documentation in README.md and SKILL.md provides examples of its usage via python3 commands.
[INDIRECT_PROMPT_INJECTION]: The skill ingests external research data which could theoretically contain malicious instructions. The risk is mitigated by treating the data as structured content and using extensive safeguards.
Ingestion points: scripts/providers/ handles API responses containing titles and abstracts from third-party academic sources.
Boundary markers: Output is strictly formatted into JSONL artifacts validated against the rls.paper.v1 schema to prevent context leakage.
Capability inventory: Subprocess calls are limited to the skill's own internal scripts; file system access is restricted to the local workspace.
Sanitization: scripts/search_runner.py includes a _safe_error function that redacts API keys and URLs from logs. Additionally, scripts/manifest.py uses safe_relative and _safe_bundle_dir to enforce strict path boundaries and prevent traversal.

research-literature-search