Skip to content

feat: allow per-task model tiering instead of forcing opus#34

Closed
mythkiven wants to merge 1 commit into
revfactory:mainfrom
mythkiven:feat/model-tiering
Closed

feat: allow per-task model tiering instead of forcing opus#34
mythkiven wants to merge 1 commit into
revfactory:mainfrom
mythkiven:feat/model-tiering

Conversation

@mythkiven

Copy link
Copy Markdown

Summary

Replaces the hard "all agents must use opus" rule with a two-tier model policy: reasoning tasks → opus, mechanical/IO tasks → sonnet.

Motivation

Closes #29 — forcing opus on log parsers and deploy runners adds cost/latency without quality benefit, and the audit checklist reverts user overrides on sync.

Scope of change

  • Skill / meta-skill logic — skills/harness/SKILL.md
  • Agent template(s) — agent-design-patterns.md, orchestrator-template.md
  • Plugin manifest
  • Documentation (README/docs)
  • CHANGELOG.md
  • CI / GitHub Actions
  • Tests

Tests

  • Manual: grep confirms no remaining "always opus" / model: "opus" mandatory wording in the three touched files.

CHANGELOG update

Y

SemVer impact

minor — new configurable behavior; existing harnesses keep working; default tier for complex roles remains opus.

Made with Cursor

Replace "always opus" with reasoning (opus) vs mechanical (sonnet)
tiers across SKILL.md, agent-design-patterns, and orchestrator template.
Closes #29.

Co-authored-by: Cursor <cursoragent@cursor.com>
@mythkiven mythkiven closed this by deleting the head repository Jun 15, 2026
@mythkiven

Copy link
Copy Markdown
Author

Superseded by #40 — branch restored, PR re-created for genuine contribution. Please review #40 instead.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[feat] Allow per-task model tiering instead of forcing opus on all agents

1 participant