[Upstream #204] 完全无法使用,没法配置第三方模型

[quangdang46]

Mirrored from upstream 1jehuang/jcode — Issue #204 by @DuanPeng1314
Original state: open
Created: 2026-05-12T19:30:44Z · Updated: 2026-05-14T10:12:11Z
This issue is an auto-mirrored copy. Comments and edits here are local to quangdang46/jcode — do not expect them to propagate upstream.

---

🛑 Auto-poke stopped because the last request failed with a non-retrya
ble error. Fix the request/session, then run /poke again if you want

[quangdang46]

Mirrored upstream Chinese-language complaint: "完全无法使用,没法配置第三方模型" ("completely unusable, can't configure third-party models"). No reproduction details. Third-party model config is documented in OAUTH.md and the README's "Self-hosted OpenAI-compatible endpoints" section. Closing as not planned — please file a fresh issue with the specific endpoint + error if it still reproduces.

[quangdang46]

Issue body is in Chinese: "完全无法使用,没法配置第三方模型" — "Completely unusable, cannot configure third-party models".

Likely covered by:

• PR docs: configuration files reference for offline / self-hosted use (#48) #226 (config reference doc) — explains how to configure custom OpenAI-compatible providers.
• PR fix(provider/minimax): JCODE_MINIMAX_REGION override for global sk-cp- keys #237 (MiniMax region override) — addresses sk-cp- key routing.
• PR feat(storage): JCODE_USE_XDG opt-in for XDG Base Directory paths (#123) #225 (XDG support).
• jcode provider add for custom endpoints.

Marking as needs-info — please re-open with specific provider + error if the above PRs do not resolve.

[quangdang46]

Verify-closing — third-party model configuration is fully supported in master:

• Built-in custom OpenAI-compatible provider: jcode login --provider openai-compatible
• One-shot setup: jcode provider add my-api --base-url https://... --model my-model --api-key-stdin --set-default
• Recent additions: GitLab Duo (PR feat(provider): GitLab Duo as OpenAI-compatible profile (#19) #263), Vertex AI (PR feat(provider): Vertex AI provider profile (#119) #283)
• Self-hosted endpoints: edit ~/.jcode/config.toml directly or use env file at ~/.config/jcode/openai-compatible.env

If specific provider configuration still fails, please file with: which provider, full error message, and jcode doctor --format json output.