OpenCode Go Model Picker

为自己的智能体挑一个 OpenCode Go 里的模型，是件麻烦事：价格、月度额度、可用模型一直在变，而「最合适」的那个又取决于你更在意省钱还是更在意能力。这个技能就是替你做完这份功课的：它会读你的 OpenCode 智能体配置、查当前的 Go 套餐，然后给你每个自定义智能体推荐一个模型——如果你的配置支持，还会附上一条回退链。整个过程只读，在你点头之前什么都不改。

Choosing which OpenCode Go model each of your agents should use is fiddly. Prices, monthly caps, and the list of available models change constantly, and the “best” pick depends on whether you care more about cost or capability. This skill does that homework for you: it reads your OpenCode agent config, checks the current Go plan, and suggests a model for each of your custom agents — plus a fallback chain when your setup supports one. It is read-only, so nothing changes until you say yes.

产出Outputs

每次运行后，它会给你一份「三段正文 + 数据附录」的报告：

Each run ends with a three-part body plus a data appendix:

• 推荐表：每个自定义智能体一行——名称、职责介绍（一行白话说明该智能体的职责，取自其 description，没有时用已知角色基线，并按你的语言渲染）、能力侧重（trait）、当前模型、推荐模型，以及为什么（成本 / 吞吐 / 能力理由），限时、地理或训练相关的注意事项用 ⚠ 标出。
• Recommendation table: one row per custom agent — name, a one-line role overview (what the agent does, from its description or the known-role baseline when it has none, rendered in your language), capability focus (trait), current model, recommended model, and the why (a cost, throughput, or capability reason), with ⚠ markers for limited-time, geo, or training caveats.
• 可粘贴配置：针对每个检测到的来源给出一段完整、可直接复制粘贴的 JSONC（slim 预设对应 oh-my-opencode-slim.json，原生 agent 块对应 opencode.jsonc），来源支持回退链的会保留链路。
• Paste-ready config: a complete copy-pasteable JSONC block per detected source (oh-my-opencode-slim.json for the slim preset, opencode.jsonc for a native agent block), with fallback chains preserved where the source supports them.
• 要点说明：一段话讲清为什么这么改——模式与理由、价格上限、完整的注意事项原文，以及与上次快照相比的套餐变化。
• Highlights: one paragraph explaining why the changes — mode + rationale, price ceiling, full caveat text, and plan changes vs the last snapshot.
• 数据附录：来源与抓取日期（种子标注）、快照差异、相关模型价格表、首次使用的决策因素表、待人工核实清单，以及验证命令。
• Data appendix: sources with fetch dates (seed labeling), the snapshot diff, the relevant-models price table, the first-run decision-factors table, the manual-verification list, and the verify commands.

报告之后它会停下来问你，确认后才应用改动。

It then stops and asks before applying anything.

特点Features

• 只读：在你确认之前绝不写任何配置，改动会先完整预览。
• Read-only: never writes config until you confirm; every change is previewed first.
• 实时：每次运行重新抓取 Go 套餐，不推荐已下架模型或过期价格。
• Fresh: re-fetches the Go plan every run, so no retired model and no stale price.
• 多来源：原生智能体、oh-my-opencode-slim 预设、其他插件注入的智能体都能读。
• Multi-source: native agents, oh-my-opencode-slim presets, and agents from other plugins.
• 跳过自带：OpenCode 自带的 build、plan 及内置 subagent 一律不碰。
• Built-ins skipped: OpenCode’s own build, plan, and built-in subagents are left alone.
• 带回退：支持链的工具给一条有序故障转移链，首选被限流也不中断会话。
• Fallback-ready: an ordered failover chain wherever the tool supports one, so a cap does not end the session.
• 有来源：每个数字都带来源与抓取日期，拿不准的标出，绝不编造。
• Sourced: every number carries a source and a fetch date; unknowns are flagged, never invented.
• 三种模式：budget / balanced / quality，按价格与能力的取舍来选。
• Three modes: budget / balanced / quality, trading price against capability.
• 实验室为准：能力（尤其视觉输入）以模型所属实验室的官方文档核实，不从模型名推断。
• Lab-verified: capabilities (especially vision) are checked in the model’s own lab docs, not inferred from the id.

安装Install

需要 OpenCode（支持 Agent Skills）和 Node.js 22+；运行时只用得到 SKILL.md、references/ 和 scripts/。

Requires OpenCode with Agent Skills support and Node.js 22+. At runtime only SKILL.md, references/, and scripts/ are used.

把下面这段发给 AI，它会帮你装好。

Paste the block below into your AI agent and it will install the skill for you.

帮我安装 "OpenCode Go Model Picker" 这个技能。

1. 获取仓库：https://github.com/sogeisetsu/opencode-go-model-picker
2. 只把下面这些复制到我的全局 OpenCode 技能目录
   （~/.config/opencode/skills/opencode-go-model-picker/）：
   - SKILL.md
   - references/
   - scripts/
3. 不要复制仓库里的其他内容（README、LICENSE、assets 等）。
4. 验证：在安装目录里运行 "node scripts/go.mjs --help"，确认列出四个动词。
5. 告诉我安装路径以及是否成功。

Install the "OpenCode Go Model Picker" skill for me.

1. Get the repository: https://github.com/sogeisetsu/opencode-go-model-picker
2. Copy only these into my global OpenCode skills directory
   (~/.config/opencode/skills/opencode-go-model-picker/):
   - SKILL.md
   - references/
   - scripts/
3. Do not copy the rest of the repo (README, LICENSE, assets, ...).
4. Verify by running "node scripts/go.mjs --help" inside the installed
   folder and confirming it prints the four verbs.
5. Tell me the install path and whether it worked.

或者自己装Or install it yourself

用 skills.sh 的 CLI 装到用户目录，只给 OpenCode：

Install into your user directory for OpenCode with the skills.sh CLI:

npx skills add sogeisetsu/opencode-go-model-picker -g -a opencode

也可以从最新 release 下载最小安装包（*-min.zip，包内只有 SKILL.md、references/、scripts/ 和 LICENSE），解压到 ~/.config/opencode/skills/ 即可；或者直接 clone 仓库：

Or download the minimal package from the latest release (*-min.zip: SKILL.md, references/, scripts/, and the LICENSE) and unzip it into ~/.config/opencode/skills/, or clone the repository:

# Linux / macOS
git clone https://github.com/sogeisetsu/opencode-go-model-picker.git \
  ~/.config/opencode/skills/opencode-go-model-picker

# Windows (PowerShell)
git clone https://github.com/sogeisetsu/opencode-go-model-picker.git `
  "$env:USERPROFILE\.config\opencode\skills\opencode-go-model-picker"

装好后在技能目录里运行 node scripts/go.mjs --help，能看到四个动词就说明装好了。

Then run node scripts/go.mjs --help in the skill folder; seeing the four verbs means it is installed.

快速上手Quick start

安装完成后，在 OpenCode 里直接输入斜杠命令 /opencode-go-model-picker 即可调用，不需要再输入任何内容——技能会自己去读你的智能体配置、抓取当前 Go 套餐，然后给出推荐。

Once installed, invoke the skill in OpenCode with its slash command /opencode-go-model-picker — no input needed. It reads your agent config, fetches the current Go plan, and comes back with its recommendations.

TUI 里如果它没出现在 / 的补全列表中，先输入 /skills，从技能列表里选同名项即可。

In the TUI, if it does not show up in the / autocomplete list, type /skills and pick it from the skill list.