askill
feedback-vector-loop

feedback-vector-loopSafety 95Repository

Build reusable directional-feedback loops for any creative/agentic pipeline. Use when replacing binary thumbs-up/down with structured critique vectors (more/less/avoid/replace), compiling feedback into weight adjustments, ranking candidates with those vectors, and iteratively improving outputs with deterministic artifacts.

0 stars
1.2k downloads
Updated 4/5/2026

Package Files

Loading files...
SKILL.md

Feedback Vector Loop

Use this skill to operationalize directional feedback:

  • not just "good/bad"
  • but "move output toward X and away from Y"

This pattern is reusable across brands (e.g., Jacksnaps, Digipop, any creative pipeline).

When to use

  • You have candidate outputs and human critique.
  • Binary labels are too weak for improvement.
  • You want iterative quality gains while preserving deterministic control.

Core model

Each feedback record should include:

  1. Verdict: up | revise | down
  2. Direction:
    • more[]
    • less[]
    • avoid[]
    • replace{old:new}
  3. Optional candidate context (title, slogan, rationale, etc.)

Deterministic pipeline

  1. Capture feedback log (feedback_log.jsonl).
  2. Compile vector profile (feedback_vector.json).
  3. Rank new candidates with base score + vector score.
  4. Apply replacement rules and penalties/bonuses.
  5. Emit stage trace and artifacts for auditability.

Runbook

1) Compile vector from feedback

python scripts/compile_feedback_vector.py \
  --input <feedback_log.jsonl> \
  --output <feedback_vector.json>

2) Rank candidates with vector

python scripts/rank_candidates_with_vector.py \
  --candidates <candidates.json> \
  --vector <feedback_vector.json> \
  --output <ranked_candidates.json>

3) Validate data contract

Use references/feedback-data-contract.md for required fields.

Design invariants

  • Keep feedback rows append-only.
  • Use weighted token adjustments, not opaque hidden state.
  • Persist artifacts every run (traceability).
  • Separate creative generation from deterministic scoring.

Suggested artifact set per run

  • feedback_log.jsonl
  • feedback_vector.json
  • ranked_candidates.json
  • stage_trace.jsonl
  • summary.json

Install

Download ZIP
Requires askill CLI v1.0+

AI Quality Score

67/100Analyzed 3/29/2026

Well-structured technical skill with clear directional-feedback model and runbook. Reusable across pipelines with explicit design invariants. Missing full script implementations and error guidance reduce actionability. Project-specific context evident but pattern is broadly applicable.

95
78
72
62
58

Metadata

Licenseunknown
Version-
Updated4/5/2026
Publisherdelorenj

Tags

ci-cd