Jev Agent Tools
Agent-facing, scoped Jev questions over supplied context or repository files. Inspired by IndyDevDan's "10 Levels of Jev For Agentic Engineers" cheap-read and files-at-scale demos: agents get typed Noul, Choice, and Score judgments without loading every candidate file into their conversation. One request answers many questions over shared state; bounded concurrent requests judge independent files through @swamp/typesafe-ai and @webframp/typesafe-batch. See the bundled README for live matgreten.dev examples and end-to-end timings.
Initial release: scoped agent-facing Jev questions over selected repository files, with concurrent per-file triage and one-call multi-question decisions.
Global Arguments
| Argument | Type | Description |
|---|---|---|
| jevModel | string | |
| repositoryRoot | string | Explicit repository root used to scope file reads. |
| Argument | Type | Description |
|---|---|---|
| mode | enum | |
| paths? | array | |
| globs? | array |
| Argument | Type | Description |
|---|---|---|
| preparationName | string | |
| sourceModel | string |
Resources
Prepare scoped state, ask Jev once for combined questions or concurrently across files, and return answer-only results.
- Has README or module doc2/2earned
- README has a code example1/1earned
- README is substantive1/1earned
- Most symbols documented1/1earned
- No slow types (deprecated)1/1earned
- Dependencies pass trust audit2/2earned
- Has description1/1earned
- Platform support declared (or universal)2/2earned
- License declared1/1earned
- Verified public repository2/2earned