Refine dispatch prompts and agent-selection criteria through iterative scoring cycles.
npx skills add https://github.com/jadecli/researchers --skill refine-jadecli
Please help me install this Agent Skill. Skill: refine Source: https://github.com/jadecli/researchers/tree/main/claude-multi-agent-dispatch/.claude/skills/refine Command: npx skills add https://github.com/jadecli/researchers --skill refine-jadecli
Run a refinement cycle to improve dispatch prompts or agent selection criteria by identifying gaps, generating candidate improvements, scoring them, and updating the contextual guidance to boost decision quality.
Provide your current prompts or agent-selection criteria and run refinement to generate improved prompts and update the context delta.
High-intent search queries and answers about installing and using this skill.
To refine prompts for smarter agent dispatch, you run a refinement cycle that loads prior context, generates candidate improvements via seed_improver, scores them, and updates the context delta to boost decision quality.
Prompt refinement for agent selection criteria is the process of identifying gaps in current prompts and selector configurations, then generating and evaluating candidate prompts iteratively to improve dispatch decision quality.
You improve agent dispatch prompts across iterative cycles by applying refinement to prompts and pipeline configurations, generating candidates via seed_improver, scoring them, and updating contextual guidance while detecting stagnation or regression.
Yes, prompt refinement detects stagnation or regression during iteration by updating the context delta after scoring candidate prompts, ensuring that subsequent cycles maintain progress and improve overall decision quality.
Yes, you can use seed_improver to generate and evaluate candidate prompts by loading prior context, producing improved prompts across iterations, scoring them, and selecting the best candidate to update the context delta.
arroyo2009jose-maxLoad project context from BUSINESS_LOGIC.md, features, database, and configuration.
AesopScottOptimizes sub-agent dispatch performance by tuning lane concurrency and fan-out parameters.
benmaierDesign and validate data-science agent prompts and parallel-agent configurations.
alpoxdevGenerate and optimize reusable prompt packs with output schemas and regression fixtures.
