skills.rest
delorenj
delorenjCOMMUNITY
Explore all skills by this author →
Website𝕏@CodeNDaggerGitHub

feedback-vector-loop

Compile structured critique into weighted feedback vectors for iterative output refinement.

Updated Feb 10, 2026
Search Tags:#feedback#agentic workflow#iterative improvement#structured critique#vector scoring#creative pipeline
One-click install
npx skills add https://github.com/delorenj/00_ceiling --skill feedback-vector-loop
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: feedback-vector-loop
Source: https://github.com/delorenj/00_ceiling/tree/main/feedback-vector-loop
Command: npx skills add https://github.com/delorenj/00_ceiling --skill feedback-vector-loop

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires python3, and includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill replaces vague, binary feedback (like thumbs up/down) with structured, directional critique, enabling iterative improvement of creative or agentic outputs.

Core Features & Use Cases

  • Directional Feedback: Define feedback using 'more', 'less', 'avoid', and 'replace' vectors.
  • Iterative Improvement: Compile feedback into weighted adjustments to refine outputs over time.
  • Deterministic Control: Maintain auditability and control through persistent artifacts and clear scoring.
  • Use Case: When refining marketing copy, instead of just saying "good," you can specify "more 'urgent' tone," "less 'passive' language," and "replace 'innovative' with 'cutting-edge'."

Quick Start

Use the feedback-vector-loop skill to compile feedback from feedback_log.jsonl into a vector profile.

Frequently Asked Questions about feedback-vector-loop

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I structure feedback for iterative improvement of creative pipeline outputs?▼

Structure feedback for iterative improvement using 'more', 'less', 'avoid', and 'replace' vectors. This replaces vague binary feedback with directional critique that compiles into weighted adjustments for deterministic output refinement.

What is vector scoring for agentic workflows and how does it work?▼

Vector scoring for agentic workflows operationalizes directional feedback loops by compiling structured critique from a log into weighted adjustments. It ranks candidates deterministically based on these compiled feedback vectors.

How do I compile feedback vectors from a JSONL log to refine marketing copy?▼

Compile feedback vectors from a JSONL log by running Python scripts that process directional critique like 'more urgent tone' or 'replace innovative with cutting-edge'. The script generates a vector profile for ranking output candidates.

Do I need Python3 to run feedback-vector-loop for structured critique?▼

Yes, you need Python3 installed to run feedback-vector-loop. The Skill relies on Python scripts to compile structured critique into weighted adjustments and rank candidates based on feedback vectors.

Does feedback-vector-loop support deterministic artifacts for creative pipeline auditing?▼

Feedback-vector-loop supports deterministic artifacts for creative pipeline auditing. It maintains auditability and control through persistent artifacts and clear scoring compiled from structured critique vectors.

Why use directional critique instead of binary feedback for agentic workflows?▼

Use directional critique instead of binary feedback for agentic workflows to avoid vague thumbs up/down ratings. Structured vectors like 'more', 'less', 'avoid', and 'replace' enable precise, weighted adjustments for iterative output improvement.

Related Skills

Explore More Design & Creative Skills→
shiimanshiiman
2

gitignore-check

Identify missing .gitignore patterns and propose ignore rules before commits.

Community
Intermediate
arashshahidi1997arashshahidi1997
1

questio-next

Rank unblocked research tasks by dependency depth and evidence gaps.

Community
Advanced
kunalrawat425kunalrawat425

design-shotgun

Generate multiple AI design variants and open a side-by-side comparison board for UI screens or flows.

Community
Advanced
selcukyucelselcukyucel
1

prompt-test

Run a prompt against test inputs and score outputs with a binary rubric.

Community
Advanced