/fresh-eyes — break the local minimum

SkillAI & models

Reset review context when bead polishing or code review has flatlined. Requires starting a brand-new Claude / Codex session. Use when you're seeing the same small edits round after round.

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the /fresh-eyes — break the local minimum skill

What this skill tells your AI

The instructions your AI receives, as published by hoangsonww/forge-agentic-coding-cli in .claude/skills/fresh-eyes/SKILL.md and read by ahel’s review.

When a session has accumulated too many assumptions, fresh eyes beat tired eyes. Context exhaustion can feel like convergence.

Prerequisite

Start a new Claude Code or Codex session. Don't just /clear.

Two-step prompt

Send, in order, the two prompts at .flywheel/prompts/fresh-eyes.md:

  1. First, a deep context load (AGENTS.md + CLAUDE.md + FLYWHEEL.md + README.md + code investigation).
  2. Then, a review of .beads/beads.jsonl against .flywheel/plans/<PLAN>.md.

When this operator pays off most

  • Bead polishing flatlined after 3–4 rounds.
  • Deep code review keeps finding the same class of bug.
  • A major architectural decision feels locked in but nobody remembers justifying it.

After

If the fresh session finds new issues → back to /polish-beads. If it comes back clean → you're genuinely at convergence. Proceed.

Signals

GitHub stars
22
Forks
9
Last commit
Sep 2026
Advanced
Catalog kind
skill
Gateway key
fresh-eyes-hoangsonww
Source
github.com/hoangsonww/forge-agentic-coding-cli