autodev-loop
Warn
Audited by Gen Agent Trust Hub on Mar 17, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill parses and executes arbitrary shell commands defined as "fitness functions" within the frontmatter of task entity files such as plans and chores. This allows the execution of logic provided in data files, which is a significant attack surface if the input files are untrusted. Evidence: Ingestion points (SKILL.md reads from .lev/pm/plans/), Capability (Bash), Boundary markers (Absent), Sanitization (Absent).
- [EXTERNAL_DOWNLOADS]: Orchestration relies on the "lev" package executed via npx, which involves downloading and running code from the npm registry at runtime. As these tools belong to the author's vendor namespace (lev-os), they are documented as trusted vendor resources.
- [COMMAND_EXECUTION]: Uses CronCreate to manage recurring execution ticks, providing the agent with persistence on the host system to maintain its autonomous development loop.
Audit Metadata