The Agent Skills Directory

[SAFE]: The skill identifies and processes government documents using a local Python script (scripts/document-analyzer.py). This script uses regular expressions to extract structured data like dates and reference numbers and does not perform any network calls or unsafe code execution.
[SAFE]: Instructions for looking up laws direct the agent to use WebFetch on main.knesset.gov.il, the official and trusted legislative database of the Israeli Knesset.
[SAFE]: There are no indicators of credential harvesting, obfuscation, or persistence. The requested tools (Bash, Read, Edit, WebFetch) are appropriate for the skill's functionality as a document interpretation and research tool.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection (Category 8) as it processes untrusted input from user-provided documents.
Ingestion points: SKILL.md Step 1 (reading pasted text or file content).
Boundary markers: Absent.
Capability inventory: Bash (script execution), Read (local file access), WebFetch (network access for law research).
Sanitization: Absent (the script uses regex parsing but does not sanitize for instructions). This surface is considered safe within the context of the skill's specific purpose and the absence of high-privilege execution sinks.

israeli-bureaucracy-decoder