dev-orchestrator
Warn
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: MEDIUMOBFUSCATIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [OBFUSCATION]: The file
scripts/dev-workflow.cjsis a bundled and minified JavaScript execution script. This obscures the logic of the tool and significantly hinders security analysis, as it includes multiple third-party libraries (such as js-yaml, micromatch, and gray-matter) and custom logic in a single non-human-readable format. - [INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection by ingesting untrusted data from project files.
- Ingestion points: The skill reads
checkpoint.md,00-master-plan.md, and various PRD files located within the project's.dev/directory viascripts/dev-workflow.cjs. - Boundary markers: The instructions lack explicit boundary markers or XML-style tagging for the content interpolated into worker briefs, though it instructs the agent to copy step text "verbatim."
- Capability inventory: The skill possesses the ability to execute shell commands (
bash), run Node.js scripts, and interact with theherdrCLI to start and prompt worker agents. It also executesgitcommands through its Node.js helper. - Sanitization: Path validation is performed in
scripts/validate.shto prevent traversal (e.g., checking for..), but there is no evidence of sanitization or escaping of the actual PRD content before it is sent to worker agents. - [COMMAND_EXECUTION]: The skill frequently executes local shell scripts (
scripts/discover.sh,scripts/validate.sh) and a Node.js CLI tool (scripts/dev-workflow.cjs). The latter also spawns subprocesses to interact with thegitbinary for repository state management.
Audit Metadata