mirror of https://github.com/merlinhu1/codex-game-studio.git synced 2026-08-25 07:54:34 +02:00

Files

T

7e8846a7fb feat: add deterministic model tier routing (#14 )

* feat: add deterministic model tier routing

Define Sol, Terra, and Luna policies with fixed fallbacks, explicit surface assignments, runtime precedence, and Luna escalation. Validate generated metadata and expose tier overrides across role, task, and orchestration commands.

* fix: infer model tiers from surface models

* fix: rebalance prompt model routing

* fix: decouple prompt effort routing

---------

Co-authored-by: MerlinH <merlinh221@gmail.com>

2026-07-10 18:49:20 +10:00

2.3 KiB

Raw Permalink Blame History

model, model_reasoning_effort, primary-agent, linked-skills, phase, risk, argument-hint, source-reference, source-hash, output-artifacts

model

model_reasoning_effort

primary-agent

linked-skills

phase

risk

argument-hint

source-reference

source-hash

output-artifacts

gpt-5.6-terra

high

performance-analyst

cgs-perf-profile

cgs-bugfix

review

medium

Provide a perf profile request with the objective or decision, target files/assets/milestone, scope constraints, owner or handoff needs, and required verification evidence.

.codex/workflows/perf-profile.md

e8b50d85075eea5dc75d5bd2f4a299591a615accbc074ab7307ab3d8a48e056b

plan

changed-files

verification-evidence

handoff

Perf Profile Workflow

Purpose

Plan or review performance profiling across frame time, memory, loading, asset cost, bottlenecks, and measurement evidence.

Inputs

AGENTS.md
.codex/studio.json
.codex/workflows/perf-profile.md

Role

Performance Analyst (performance-analyst) owns this workflow.

Taxonomy

Category: qa-testing CCGS adaptation coverage:

performance profiling
optimization triage

CLI aliases:

perf-profile

Outputs

Performance report
Optimization priorities
Measurement plan

Validation

Budget and metric are named
Measurement method is reproducible
Risks and trade-offs are clear

Phase Gates

Confirm project state, owner role, write policy, and selected context.
Name acceptance criteria before implementation or review work.
Do not advance to handoff until evidence is recorded or a blocker is explicit.

Required Artifacts

Summary of the workflow result.
Files, assets, tasks, or docs changed or proposed.
Verification evidence and unresolved risks.

Context Contract

Load AGENTS.md, .codex/studio.json, this workflow, the primary agent, linked skills, and only task-relevant project files.
Avoid broad context unless the user explicitly approves it.

Output Contract

Decision or change summary.
Step-by-step work performed or planned.
Evidence, blockers, warnings, and next owner.

Stop Conditions

Required project state, approval, target files, or verification path is missing.
The task crosses into another role without an explicit handoff.
The workflow would require generated prompt mirrors or hidden automation.

Handoff

Report changed files, validation evidence, residual risks, and the next owner only when ownership changes.

2.3 KiB Raw Permalink Blame History