[kixelated]

Lowers every writer one rung so xhigh is XL's alone, moves XS to Haiku, and adds a quest so the headless land fixer stops writing quest files. The user decided all of this on 2026-11-05 while spawning the token waste quests. Agents already running keep their efforts.

• Route by size: the new write lists.
  • XS: Haiku high, falling back to Sol low.
  • S, M, L, XL: Opus low, medium, high, xhigh.
  • Sol fallbacks: XS low, S medium, M high, plus a new L fallback at xhigh. XL keeps no fallback.
  • Sol reviews now follow Single-pass review's table. That decision (moq-dev/quest#509) is newer than the "reviews mirror the write list" rule from moq-dev/quest#507, which it replaces. Opus reviews follow Opus's write efforts.
• The watcher writes flaky quests [S], new in the queue watcher epic. Today the land fixer, a headless Codex agent, is the one place a sub-agent writes a quest. It writes quest/m0/flaky/<name>.md in its disable commit (assets/land-fix.md). After this quest, quest land watch writes that file from a template. Every skill already leaves quest writing to the coordinator.

Decision prompts

Chosen options are marked ✅.

• Goal: ✅ each size gets a cheaper writer, so xhigh is reserved for XL.
• Running agents: ✅ let them finish at their old efforts (rejected: restart them at the new efforts).
• XS writer: ✅ Haiku high → Sol low (rejected: Haiku medium, Opus low, Sol medium, DeepSeek through opencode). No benchmark has published Opus 5.5 at low effort. Bito puts Haiku high 3-5 points behind Opus medium, at 1/4 to 1/23 of the cost.
• Sol fallbacks: ✅ one rung lower, plus an L fallback at xhigh (rejected: unchanged).
• Reviews: ✅ the Single-pass review table wins (rejected: the write list mirrors the reviews).
• Land fixer quests: ✅ the watcher writes them from a template (rejected: a coordinator plans them later; keep the exception).
• Milestone: ✅ m0, queue watcher epic (rejected: m1).

Models

Size: S
Implementer: Claude Code, claude-opus-5-5, high effort

Validation

quest check: 77 documents ok.

(Written by Claude Opus 5.5 with high effort)

🤖 Generated with Claude Code

Review

• Reviewer: Codex, gpt-6.1-sol, medium effort
• Session: 01a1ade8-fa42-7b83-a6bf-ea72404c8b9a, round 1
• Head: 860b38b
• Base: origin/main at 73e0caf
• Findings: none

[pullfrog]

  ^{｜ Review this ➔}

[coderabbitai]

Warning

Review limit reached

You've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository.

Next included review available in 8 minutes.

Check out review usage here.

View limit details

Limit details: You’ve used the included review currently available.

Learn how review limits work.

Review configuration:

Run configuration

• Source mode: Auto
• Repository: Unavailable
• Branch: Unavailable
• File: No configuration file from this repository applied
• Commit: Unavailable
• Layers: Organization UI
• Review profile: CHILL
• Plan: Advanced
• Run ID: 19f0d231-2f5b-4043-9d62-39b40e93f4bd

Commits

Reviewing files that changed from the base of the PR and between 73e0caf and 860b38b.

Files selected for processing: 4

• quest/m0/model-routing/README.md
• quest/m0/model-routing/models.md
• quest/m0/queue-watcher/README.md
• quest/m0/queue-watcher/flaky-quest.md

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

Share

• X
• Mastodon
• Reddit
• LinkedIn

[kixelated]

Automated review by review (OpenAI)

Reviewed head: 860b38b

• [P2] Update the epic's acceptance goal with the new routing. The added revision assigns XS to Haiku, but the epic Goal still requires the live XS and S spawns to both use Opus, and says Sol falls back only through M. A correct implementation of the new table therefore fails the stated completion criterion. Update the Goal to require XS on Haiku, S on Opus, and describe fallbacks through L.

Direction: keep the ranked-routing and watcher-owned template approach; reconcile that acceptance criterion before implementing the plan. No other actionable findings in the four-file planning change.

Verification: GitHub-only inspection of the changes, referenced plans, routing/watcher code, and existing review/flaky tests. No builds or tests run. The author reports quest check passing for 77 documents; this head's Actions checks are skipped. External benchmark and pricing claims were not independently verified.