treeherder

Warn

Audited by Gen Agent Trust Hub on Feb 18, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • EXTERNAL_DOWNLOADS (MEDIUM): The skill installs treeherder-cli via cargo install --git https://github.com/padenot/treeherder-cli and executes lumberjackth via uvx. These sources are not part of the trusted organization whitelist (e.g., mozilla/).\n- REMOTE_CODE_EXECUTION (MEDIUM): Facilitates the automated download and execution of binaries and packages from external, untrusted repositories. Severity is reduced from HIGH to MEDIUM as this is the skill's primary intended purpose.\n- COMMAND_EXECUTION (LOW): Uses subprocesses to call external CLI tools to perform its primary functions.\n- DATA_EXFILTRATION (LOW): Connects to treeherder.mozilla.org to retrieve logs and artifacts. While legitimate for its purpose, it involves outbound network requests to a non-whitelisted domain.\n- PROMPT_INJECTION (LOW): Exposed to Indirect Prompt Injection (Category 8) by processing untrusted log data.\n
  • Ingestion points: treeherder-cli --fetch-logs and lj errors fetch external text data from CI logs.\n
  • Boundary markers: None identified in the provided tool usage instructions.\n
  • Capability inventory: Shell command execution capabilities via CLI tools.\n
  • Sanitization: No sanitization of external log content is described before processing.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 18, 2026, 03:51 AM