⚡ Workflows
Feedback Loop (iterative refinement)
Produce high-quality output through iterative cycles of drafting and reviewing until convergence criteria are met, without a fixed hierarchy.
Bundle files
📄ORCHESTRATION.md— Orchestration
1# ORCHESTRATION.md
2
3## Diagram
4
5 ┌──────────────────────────────────────────┐
6 │ Initial Input │
7 └────────────────────┬─────────────────────┘
8 ▼
9 ┌──────────────────┐
10 │ Draft Agent │
11 │ (produce v1) │
12 └────────┬─────────┘
13 ▼
14 ┌──────────────────┐
15 │ Review Agent │
16 │ (evaluate + score│
17 │ against rubric) │
18 └────────┬─────────┘
19 ▼
20 ┌──────────┐
21 │Converged?│
22 └────┬─────┘
23 no │ yes
24 ┌─────────────┴─────────────┐
25 ▼ ▼
26 ┌──────────────┐ ┌──────────────────┐
27 │ Feedback │ │ Final Delivery │
28 │ (specific │ └──────────────────┘
29 │ fixes) │
30 └──────┬───────┘
31 │
32 ▼
33 ┌──────────────┐
34 │ Draft Agent │
35 │ (revise) │──── loop back to Review
36 └──────────────┘
37
38
39## Workflow
40Feedback Loop (iterative refinement)
41
42## Objective
43Produce high-quality output through iterative cycles of drafting and reviewing until convergence criteria are met, without a fixed hierarchy.
44
45## Roles
46- **Draft Agent:** produces and revises work based on feedback.
47- **Review Agent:** evaluates output against a rubric, scores quality, provides actionable feedback.
48- **Convergence Monitor:** tracks iteration count and quality scores to prevent infinite loops.
49
50## When to use
51- Creative or analytical output that benefits from refinement (docs, code, proposals).
52- When quality is hard to get right on the first attempt.
53- Tasks where incremental improvement is more effective than a single pass.
54
55## Protocol
561. Draft Agent produces initial version (v1).
572. Review Agent evaluates against rubric and assigns a quality score (0–100).
583. If score ≥ threshold (configurable, default 85) → deliver.
594. If score < threshold → Review Agent provides specific, actionable feedback.
605. Draft Agent revises based on feedback → produces v(n+1).
616. Repeat from step 2.
627. If max iterations reached without convergence → deliver best version with quality disclaimer.
63
64## Convergence criteria
65- **Quality threshold:** 85/100 (configurable)
66- **Max iterations:** 5 (prevents infinite loops)
67- **Improvement floor:** if score improves < 3 points between iterations, stop early (diminishing returns)
68
69## Rules
70- Feedback must be specific and actionable (no vague "make it better").
71- Each iteration must reference what changed from previous version.
72- Review Agent must not rewrite — only evaluate and guide.
73- All versions preserved for comparison.
74
75## Deliverables
76- Final output (best version)
77- Iteration log (version, score, feedback summary)
78- Convergence report (iterations used, score trajectory)
79
Lines: 79 | Words: 347
Install
Copy bundle to your OpenClaw workspace.
curl -fsSL https://raw.githubusercontent.com/cerealskill/openclaw-agents/main/install.sh | bash -s workflow feedback-loop ENRate this agent
Loading...
Sign in to rate this agent
Includes
- ✓ ORCHESTRATION.md
Info
- Slug
- feedback-loop
- Lines
- 79
- Words
- 347