http-service-patterns

Pass

Audited by Gen Agent Trust Hub on Sep 5, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • Command Execution and Process Management: The skill uses subprocess.run to call lsof and os.kill to manage process lifecycles. This pattern is designed to prevent port-binding failures during service restarts. While functional, it involves system-level interactions that should be scoped to specific ports and processes to ensure they do not affect unrelated system components.- Disabled SSL Verification: The pattern for background poll loops uses httpx.AsyncClient(verify=False). As noted in the skill's documentation, this is intended to support connections to upstream services using self-signed certificates. This configuration bypasses standard certificate validation and should be used only in trusted internal network environments.- Indirect Prompt Injection Surface: The skill defines endpoints for WebSocket relaying and Server-Sent Events (SSE) that ingest and process external data.
  • Ingestion points: Data enters the context via WebSocket messages through websocket.receive and by reading from external log files like events.jsonl.
  • Capability inventory: The skill patterns include file system access via open, network operations using websockets.connect, and process management with subprocess.run.
  • Boundary markers: The provided snippets do not include explicit boundary markers or instructions to ignore embedded commands in the relayed data stream.
  • Sanitization: The snippets show instance_id being interpolated into file paths without explicit validation or sanitization, which is a common implementation area requiring additional safety logic in production code.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 5, 2026, 04:27 AM
Security Audit — agent-trust-hub — http-service-patterns