experiment-queue

Warn

Audited by Gen Agent Trust Hub on Nov 12, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The scheduler script scripts/queue_manager.py constructs shell commands for execution on remote hosts using subprocess.run(shell=True) without consistent sanitization of all inputs.
  • Evidence: The run function in scripts/queue_manager.py is a thin wrapper around subprocess.run(cmd, shell=True), which is used for multiple sensitive operations.
  • Evidence: In launch_job, several variables derived from the user-provided manifest, such as job['id'] and conda_env, are interpolated into command strings without sufficient sanitization. For example, screen_name = f"EQ_{job['id']}" is used directly in screen -dmS {screen_name} without quoting, allowing for command injection if job['id'] contains shell metacharacters.
  • [INDIRECT_PROMPT_INJECTION]: The skill's primary workflow involves processing external data (grid specifications and manifests) to automate experiment execution, which creates an entry point for malicious instructions.
  • Ingestion points: scripts/build_manifest.py reads user-provided YAML/JSON files, and SKILL.md instructions guide the agent to parse natural language descriptions into a manifest.
  • Boundary markers: Absent. There are no explicit delimiters or instructions to ignore embedded commands within the processed configuration data.
  • Capability inventory: The skill possesses the capability to execute arbitrary shell commands on remote servers, write files, and perform network operations via ssh and scp.
  • Sanitization: Partial. While shlex.quote is used for some path-related variables, other fields in the manifest are used in shell commands without proper escaping.
  • [DYNAMIC_EXECUTION]: The skill dynamically generates and executes complex shell command strings at runtime, increasing the risk of injection vulnerabilities.
  • Evidence: The launch_job function in scripts/queue_manager.py assembles a multi-line bash script (including directory changes, environment activation, and environment variable setting) which is then passed to screen and bash -c for execution.
  • [EXTERNAL_DOWNLOADS]: The skill automates the transfer and execution of code on remote servers as part of its core functionality.
  • Evidence: Step 3c in SKILL.md uses scp to copy helper scripts queue_manager.py and build_manifest.py to the target server, which are subsequently executed via ssh in Step 3d.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Nov 12, 2026, 06:08 AM