The Agent Skills Directory

[Prompt Injection] (SAFE): No instructions to bypass safety filters, extract system prompts, or disregard instructions were found. The skill emphasizes grounding in specific reference files.
[Data Exposure & Exfiltration] (SAFE): No hardcoded credentials (API keys, tokens), sensitive path access (SSH/AWS), or suspicious network operations were detected.
[Indirect Prompt Injection] (LOW): The skill is designed to process user-provided graph schemas and Cypher queries. While it lacks explicit XML-style boundary markers for untrusted input, it mitigates injection risks by implementing a strict 'Validations' system that specifically flags unparameterized queries (f-strings) and other unsafe query constructions as errors.
[Remote Code Execution] (SAFE): No patterns of remote script downloading, execution via shell piping, or use of untrusted external dependencies were identified.
[Dynamic Execution] (SAFE): The provided Python code snippets in patterns.md are instructional examples for data modeling and do not utilize eval(), exec(), or other dynamic execution sinks on untrusted input.
[Command Execution] (SAFE): No instances of subprocess.run, os.system, or similar system call capabilities were found in the skill logic.

graph-engineer