Execute multiple Algorithm cycles with iteration-to-iteration learning and review.
npx skills add https://github.com/davdunc/pai-framework --skill loop-davdunc
Please help me install this Agent Skill. Skill: Loop Source: https://github.com/davdunc/pai-framework/tree/main/skills/Loop Command: npx skills add https://github.com/davdunc/pai-framework --skill loop-davdunc
It solves the problem of getting a better result when one execution is not enough, by running multiple full Algorithm cycles that build on prior learnings and are reviewed between iterations.
Run iterative improvement on a specific file by invoking /loop with your target path, a clear goal for what “better” means, and a chosen number of iterations.
High-intent search queries and answers about installing and using this skill.
Iterative documentation refinement is achieved by executing multiple full Algorithm cycles that build on prior learnings and apply evolving evaluation criteria between each pass.
Automated research output improvement uses an opt-in autoresearch loop with dead-ends tracking and confidence gating to avoid repeatedly retrying failed approaches across iterations.
Workflow routing with human review requires setting the mode to loop, defining a clear goal for improvement, and tracking iteration history alongside evolving ISC criteria.
Yes, iterative improvement cycles can refine skill workflows by applying the Observe, Think, Plan, Build, Execute, Verify, and Learn phases repeatedly until the workflow improves.
Yes, the optional autoresearch behavior implements MAD-based confidence gating and maintains a dead-ends ledger to prevent the algorithm from repeatedly retrying failed approaches.
Evolving evaluation criteria are required because each iteration refines what the ideal outcome means, allowing the system to track cumulative improvements across the same underlying target.
andrew-starosciakRefine subagent context through a four-stage iterative retrieval loop.
danielmiesslerRefine a target through multiple Algorithm cycles with human review or autonomous iteration.
starlink-awakenAutomates iterative Algorithm cycles with human review between each pass.
neyrojasjEvaluate code against QA checklists and emit PASS/FAIL verdicts.
