The Agent Skills Directory

[SAFE]: The skill is a local verification tool designed to prevent 'hallucinations' in legal documents by performing mechanical string-matching between claims and source texts.
[COMMAND_EXECUTION]: The skill uses a Node.js script (scripts/ground-citations.mjs) to process JSON tasks. This script is self-contained, has no external dependencies, and performs only text normalization and matching operations.
[EXTERNAL_DOWNLOADS]: No external downloads or remote code executions were detected. The skill specifically highlights its 'zero-cloud' and 'RODO-safe' (GDPR-safe) design, operating strictly on local files.
[PROMPT_INJECTION]: The instructions do not contain any patterns attempting to override agent behavior or safety filters. Instead, it provides a 'generative contract' that instructs the agent to be honest about its sources and avoid inventing citations.
[DATA_EXFILTRATION]: All data processing is performed locally. The allowed-tools list (Bash, Read, Grep) is scoped appropriately for local file analysis, and the pii-egress metadata is set to none.

citation-grounding-pl