harness-migration
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONPRIVILEGE_ESCALATION
Full Analysis
- [REMOTE_CODE_EXECUTION]: Fetches and executes source code from the author's GitHub repository (
FairladyZ625/harness-anything) and installs dependencies vianpmto perform the migration tasks.\n- [PRIVILEGE_ESCALATION]: Instructs the agent to perform global software installation usingnpm install -g .to make the migration tool available system-wide.\n- [COMMAND_EXECUTION]: Executes numerous shell commands for repository initialization, git state management, and Harness daemon configuration.\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process legacy ledger files, which are untrusted external inputs. This represents an attack surface where maliciously crafted data could attempt to influence the agent's behavior during migration.\n - Ingestion points: Reads files from user-specified local directories.\n
- Boundary markers: Implements a mandatory dry-run validation step to inspect changes before final application.\n
- Capability inventory: Performs file writes, git operations, and daemon control.\n
- Sanitization: Relies on the migration tool's internal validation checks during the dry-run phase.\n- [DYNAMIC_EXECUTION]: Assembles and executes temporary shell scripts (
apply.sh) and runs TypeScript entry points directly using the Node.js runtime to manage long-running migration tasks across shell sessions.
Audit Metadata