stelow-workflow-scope-executor
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill reads external plan files (
spec-tech.md) and incorporates their content (objectives, criteria, and verification commands) directly into prompts for subagents. - Ingestion points:
spec-tech.md,stelow.json. - Boundary markers: Uses template headers (e.g.,
Objective:,Acceptance Criteria:) but lacks strict sanitization of the embedded content. - Capability inventory: Access to the
subagenttool, filesystem write access vianode, and shell command execution. - Sanitization: Minimal; data is interpolated directly into task descriptions for subagents.
- [DYNAMIC_EXECUTION]: The skill generates and executes JavaScript code via
node -eto manipulate thestelow.jsontracking file and perform data synchronization. - [DYNAMIC_EXECUTION]: The
scope-init-fallback.mdscript usesrequire()to dynamically load modules from paths constructed at runtime, including paths influenced by environment variables likeSTELOW_PARSER_PATHandSTELOW_PACKAGE_ROOT. - [COMMAND_EXECUTION]: The skill executes various shell commands for git operations (rev-parse, diff), lock management (sha1sum, date), and running verification tools specified in the project plan.
Audit Metadata