opencode-oauth-context-cap
Skill4 starsChanged 4 days agoMITView on GitHub
排查 OpenCode V2 ChatGPT OAuth 下 OpenAI 模型有效上下文被限制为 400k/272k;区分官方规格、客户端目录和实际端点能力。Triggers: GPT-6 Sol 百万上下文, 1M context, 上下文 54%, 400k, 272k, 自动压缩, ChatGPT OAuth, Codex OAuth, /api/model, limit.input, limit.context
In this repository15 other files it gives its agents.
- AGENTS.mdAGENTS.md
- dsh/AGENTS.mdAGENTS.md
- android-apk-download-verifySkill
- auto-skill-creatorSkill
- docs-stale-auditSkill
- dsh-turn-failure-diagnosisSkill
- git-branch-consolidationSkill
- opencode-share-kitSkill
- phased-refactor-and-reportSkill
- plan-reviewSkill
- powershell-utf8-safetySkill
- quality-baselineSkill
- test-verificationSkill
- white-bear-safe-rulesSkill
- windows-openssh-serverSkill
The file
Files
OpenCode OAuth 上下文限额排查
<role>确认当前会话为何提前压缩,区分模型官方窗口、OpenCode 生效目录与认证通道的真实可用窗口。</role>
Steps
- 在当前项目运行
opencode api get /api/model,只提取目标providerID/id/limit;可用opencode api get /api/session/<会话ID>只提取data.model,核对会话实际模型。 - 用
opencode auth list --format json只查看目标提供商connections[].method;不输出凭据、账户信息或原始响应。 - 对照官方模型规格:GPT-6 Sol API 文档给出
context=1050000/input=922000/output=128000。本机 OpenCode 2.0.18 以 ChatGPT OAuth 连接时,openai/gpt-6-sol生效目录是400000/272000/128000,所以 215k / 400k ≈ 54%,约 270k 输入时自动压缩。见 OpenCode #47646。 - 若目标是确切的百万窗口,先查询当前 OpenCode 版本及该连接的目录是否已有修复;独立计费的 OpenAI API key 是另一种通道,需用户同意后再切换、检查目录和实测。切勿仅凭官方 API 规格宣称 OAuth 已开启百万上下文。
Key Commands / Patterns
opencode --version
opencode api get /api/model
opencode auth list --format jsonNotes / Pitfalls
- OpenCode 的
/api/model是客户端生效元数据,不是 OAuth 端点硬限额的直接证明。OpenCode #44821 报告单个 GPT-5.6 Sol OAuth 账户/路由接受约 921k 输入;这不能推断本机 GPT-6 Sol 也必然如此。 providers.openai.models.<id>.limit是 OpenCode 配置元数据;OAuth 变换可能再次覆盖它。只改数字可能只改变显示/压缩时机,甚至因超出端点能力导致失败;改动后必须重新检查/api/model,不能把数值变大当作端点验证。- OpenCode V2 配置见 模型文档 和 压缩文档;模型规格见 OpenAI GPT-6 Sol。
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

