SkillRanker
Rank agent skills for the next step, including the option to choose none.
What Jev decides
Which skill fits best, whether each candidate fits, or whether none should be suggested.
SkillRanker uses Jev to compare visible skills against the current session, rerank a shortlist, and abstain when no skill fits.
- Jev role
- Judge which visible skill best fits the agent's next step.
- Input
- The latest request, recent session evidence, workspace signals, and skill descriptions or excerpts.
- Output
- Choice rankings and Noul fit probabilities.
- Decision frequency
- When an agent asks for a recommendation for its next step.
- Fallback
- Explicit skill requests resolve locally; failed recommendations do not block the agent.
How it works
Overview
SkillRanker is a Rust CLI that reads the current request, recent session context, workspace signals, and skills visible to the agent. Jev first compares candidates broadly, then evaluates a shortlist with richer excerpts. Choice and Noul answers support ranking and a real none-of-these outcome. The result is advisory; the agent still decides what to load.
Large inventories may be narrowed by local retrieval before Jev sees them. Explicitly requested skills are resolved locally. Ranking requires a TypeSafe API key, while offline fixtures and readiness checks can run without sharing a session.
- Language
- Rust
- License
- LicenseRef-MIT-OpenAI-Anthropic-Rider
- Author
- Dicklesworthstone
Sources
Related projects
Jev Review
A code-review pipeline assembled from small typed judgments instead of one long reviewer prompt.
What Jev decides
What is risky, which files and evidence matter, what could fail, how severe it is, and who should review it.
Distill
Route coding-agent work, reasoning effort, and context retention through bounded Jev decisions.
What Jev decides
Which configured model and effort should handle a call, what evidence deserves attention, whether another attempt is useful, and what context should be retained.
jev-model-router
Route Claude Code model tiers and reasoning effort with one parallel Jev classification.
What Jev decides
Which fast, balanced, or deep tier fits the work, how much reasoning it needs, and whether production risk requires escalation.