The Agent Skills Directory

[SAFE]: No malicious patterns such as prompt injection, data exfiltration, or unauthorized remote code execution were detected. The skill's functionality aligns with its stated purpose of diagram management.
[COMMAND_EXECUTION]: The skill instructions include running a bundled Python script for validating Excalidraw scenes.
Evidence: SKILL.md and references/excalidraw-format.md refer to executing python3 ../scripts/check_scene.py to verify the structure of generated JSON.
Context: The script is local, part of the skill package, and uses only standard Python libraries (json, argparse, math) to perform structural checks. It does not perform network operations or access sensitive system files.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it processes untrusted diagram source code provided by users or external files.
Ingestion points: User-provided diagram source in Mermaid (text), draw.io (XML), and Excalidraw (JSON) formats.
Boundary markers: The skill explicitly instructs to check the "native structure separately from the rendered composition" and uses dedicated format adapters.
Capability inventory: The skill can read local files and execute a bundled validation script. It does not have high-privilege capabilities like network access or arbitrary system command execution.
Sanitization: The check_scene.py script implements strict schema validation for Excalidraw JSON, verifying element types, IDs, and numeric properties, which mitigates the risk of processing malformed or malicious data structures.

x9-diagrams