ln-1000-pipeline-orchestrator

Pass

Audited by Gen Agent Trust Hub on Apr 25, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTIONREMOTE_CODE_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The orchestrator drives the pipeline through multiple stages by executing shell commands via a CLI utility (scripts/cli.mjs) and spawning a PowerShell process with -ExecutionPolicy Bypass to prevent the host system from sleeping.
  • [EXTERNAL_DOWNLOADS]: The skill instructs the agent to fetch configuration and reference files from the author's official GitHub repository (raw.githubusercontent.com/levnikolaevich/claude-code-skills) if they are not present in the local environment.
  • [PROMPT_INJECTION]: The skill processes untrusted external data from kanban boards and story descriptions. Ingestion points: docs/tasks/kanban_board.md and project CLAUDE.md. Boundary markers: Uses ORCHESTRATOR_BRIEF markers for some sections but lacks comprehensive isolation for all parsed task data. Capability inventory: Significant shell execution and file system access capabilities across all scripts. Sanitization: No explicit sanitization or validation of the ingested text before it is used to influence agent decisions.
  • [REMOTE_CODE_EXECUTION]: The architecture snapshot tool (scripts/lib/arch-snapshot.mjs) utilizes dynamic import() to load code from paths relative to the project root, which could lead to arbitrary code execution if the file system layout is compromised.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 25, 2026, 07:37 PM