jeo-code 0.8.32 → 0.8.34

package/CHANGELOG.md

@@ -6,6 +6,43 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 The README mirrors the latest 5 entries — regenerate with `bun run changelog:sync`.
 
+## [0.8.34] - 2026-07-21
+_Provider/model-catalog sync against gjc's actual current source (a shallow clone of the real, public `Yeachan-Heo/gajae-code` repo at v0.11.5 — direct `packages/ai/src/models.json` comparison, not inference from release notes). The system prompt's philosophy, the `/route` command, and the `tencent` provider were explicitly out of scope and are unchanged._
+
+### Added
+- **`deepinfra` provider** (`src/ai/providers/openai-compatible-catalog.ts`) — `https://api.deepinfra.com/v1/openai`, `DEEPINFRA_API_KEY`, default `deepinfra/deepseek-ai/DeepSeek-V3.2`, plus known-model picker fallbacks (DeepSeek-R1-0528, Llama-3.3-70B-Instruct-Turbo, Kimi-K2.6, Qwen3-Coder-480B, gpt-oss-120b) — a real, popular inference cloud gjc's catalog has that jeo's didn't; one data-driven catalog row, auto-registered (no new adapter code, per this file's own "add a provider = one table row" design).
+- **`litellm` provider** — `http://localhost:4000/v1` (LiteLLM's own documented local-proxy default, expected to be overridden via `jeo provider add litellm --base-url <your-proxy>`), `LITELLM_API_KEY`. Deliberately has no `knownModels` list — unlike every other row, LiteLLM is a self-hosted proxy whose model catalog is entirely user-defined (whatever their own `config.yaml` routes to), so there is no globally-fixed list to offer.
+- `src/ai/types.ts`'s `ProviderName` and `src/auth/storage.ts`'s `AuthProvider`/`API_KEY_ONLY_PROVIDERS` gained both new names; `test/deepinfra-litellm-provider.test.ts` (+5 tests: registration, env var, base URL, routing, credential-gated readiness).
+
+### Scope notes (why the rest of gjc's current provider list wasn't ported wholesale)
+- **`tencent` — explicitly kept**, per instruction, even though gjc's current `models.json` has no such entry at all (jeo's Tencent Cloud MaaS integration predates and is independent of gjc's own catalog).
+- **`moonshot`** and **`firepass`** are gjc's OWN renamed/split entries for base URLs jeo already serves under different names — `moonshot`'s `https://api.moonshot.ai/v1` is jeo's existing `kimi` provider's base URL (jeo intentionally keeps ONE `kimi` provider spanning both the OAuth Kimi-Code subscription and the plain API-key path, rather than gjc's two separate provider identifiers); `firepass`'s `https://api.fireworks.ai/inference/v1` is byte-identical to jeo's existing `fireworks` entry. Already at parity, no action needed.
+- **`opencode` / `opencode-go` / `opencode-zen`** — these are the competing `opencode` CLI's own hosted proxy/gateway products; a jeo user would need an opencode account to use them, and integrating a rival agent's own product as an upstream provider has little practical value for jeo's userbase. Not ported.
+- **`azure-openai`, `amazon-bedrock`, `google-vertex`, `cursor`, `github-copilot`, `gitlab-duo`** — each requires a fundamentally new, non-trivial auth/wire-protocol adapter (Azure resource/deployment config, AWS SigV4 request signing, GCP OAuth + region/project config, Cursor's proprietary undocumented agent API, a GitHub Copilot-specific device-auth flow, GitLab OAuth) — none fit the existing "OpenAI-compatible base URL + API key" pattern every other row here uses; each is a real, separate feature in its own right, not a catalog row. Not ported in this pass to avoid fabricating auth flows without live verification.
+- **`cloudflare-ai-gateway` / `vercel-ai-gateway`** — both require a per-account/per-gateway URL embedded in the base (`.../v1/<account>/<gateway>/...`), which isn't a good fit for a fixed catalog row; already achievable today via jeo's existing `jeo provider add --base-url` custom-endpoint mechanism.
+- **`ollama-cloud`** — `src/ai/providers/ollama.ts`'s adapter never sends an `Authorization` header (by design, for the keyless local daemon); serving the hosted `https://ollama.com` cloud variant needs that adapter to accept a bearer credential, a real (if modest) code change beyond a catalog row. Deferred as a follow-up rather than rushed.
+- System prompt philosophy (`src/prompts/`) and the `/route` command (`src/commands/launch/route-slash.ts`, `src/agent/route-history.ts`) were not touched — out of scope per the request.
+
+### Verified
+- `test/deepinfra-litellm-provider.test.ts` — 5 pass (new).
+- `bun run typecheck` clean; full `bun test` 3102 pass / 0 fail across 305 files (an initial run under heavy concurrent system load hit many unrelated timeouts across subsystems this pass never touched — browser, daemon, tokenizer, tmux, search-gitignore — all confirmed pre-existing/load-induced via isolated re-runs, then a clean full re-run came back 0 fail).
+
+## [0.8.33] - 2026-07-20
+_Real terminal-resize testing (repeated tmux resize-window + keystroke reproductions, byte-for-byte ANSI replay outside jeo/tmux to isolate root causes deterministically) uncovered and closed a progressive screen-corruption bug: resizing down to a narrow terminal (e.g. a ~20-column tmux pane) produced a growing stack of duplicate status-bar lines on every subsequent keystroke, and a separate, independently-reproducible scroll bug in the mid-turn live renderer._
+
+### Fixed
+- **Narrow-terminal overflow across three width-floor call sites, the actual root cause of the live-reproduced corruption** — `src/tui/components/status.ts`'s `renderStatusBar` and its internal `leftBudget`, and `src/tui/components/input-box.ts`'s `renderInputFrame`, each clamped their `cols` UP to a fixed floor (24, 24, 8) regardless of the REAL terminal width, so a genuinely narrow terminal (e.g. 20 columns) still received a 24-column-wide status bar / input box. A real terminal hard-wraps that overflow, which desynced the idle-prompt footer's own row bookkeeping on every subsequent repaint — reproduced live via tmux (`resize-window` down + repeated keystrokes) as a stack of duplicate `⬢ model … ▦ ctx%` lines that grew by one on every keystroke, and confirmed via a byte-for-byte replay of the exact captured ANSI sequences through `cat` into a fresh tmux pane with jeo entirely absent — proving the bug lived in the emitted bytes themselves, not timing or readline interaction. All three floors now clamp to a minimum of 1, degrading gracefully (a small, correctly-sized box/bar) instead of overflowing.
+- **`src/tui/components/welcome.ts`'s narrow (`cols < 30`) one-line fallback no longer overflows** — an unbounded `jeo v{version} · {model}` line previously overflowed a real narrow terminal by up to 47 columns with a realistic provider-qualified model id (e.g. `antigravity/claude-sonnet-4-6 (antigravity)`). Now truncated to `cols - 1` — the SAME "leave the last column free" margin the boxed banner already uses elsewhere in this file, deliberately NOT `cols` exactly: a line filled to the exact terminal width followed by its own trailing newline is ambiguous to real terminals (pending-autowrap vs. explicit-LF), which independently reproduced the same duplicate-line corruption even after the overflow itself was fixed.
+- **`src/tui/renderer.ts`'s reserve-mode resize path no longer scrolls the terminal on a shrink** — `Renderer.render()`'s resize branch called `clear()` unconditionally, which zeroes `coverRows`/`prev` as an independent write; the SAME call's own reserve-block then read that now-wrong occupancy (0) and spuriously inserted real `"\n"` characters even when the post-resize frame was SHORTER than the pre-resize one. Literal `"\n"` genuinely scrolls a real terminal, corrupting whatever sits above the live frame and permanently desyncing the renderer's row math. Reserve mode now uses the same `reset()`-style bookkeeping the periodic self-heal resync already relies on (remembers the occupied rows via `coverRows`, no output) instead of an immediate `clear()`; non-reserve (alt-screen/pipe) mode is unaffected and keeps its defensive full clear. Reproduced deterministically with a standalone `Renderer` unit harness (no tmux/timing involved) before and after the fix.
+
+### Verified
+- `test/status-bar.test.ts`, `test/input-box.test.ts` — new regression tests: neither ever emits a line wider than a realistically narrow `cols` (5 through 23), closing the exact floor-overflow class found live.
+- `test/tui-welcome.test.ts` — updated + new tests: the narrow fallback is now `cols - 1`-capped (never overflows, even with a long provider-qualified model id).
+- `test/tui-renderer.test.ts` — new regression test: a reserve-mode resize down to a shorter frame never emits `"\n"`, and a second render right after stays clean (no lingering desync).
+- Full live tmux reproduction cycle for BOTH the idle-prompt path and the mid-turn `LaunchTui` path: resized a real `jeo --tmux` session down (e.g. 90×26 → 55×18, 80×24 → 20×10) and typed multiple keystrokes after each — before this pass, the idle-prompt footer visibly duplicated once per keystroke; after, it stays clean across repeated keystrokes at both a moderate (55×18) and an extreme (20×10) narrow width, confirmed via direct pane capture, not just unit assertions.
+- Full `bun test` — 3097 pass / 0 fail across 304 files.
+- `bun run typecheck` — no errors.
+
 ## [0.8.32] - 2026-07-20
 _gjc parity: TUI display-width/wrapping is now grapheme-cluster-aware for emoji sequences (VS16 presentation, skin-tone modifiers, keycaps, ZWJ-joined emoji like family/profession glyphs), fixing box-border and wrap misalignment ("깨짐") that the old per-code-point width summation produced whenever such a sequence appeared in a message — Korean/CJK-only text was already correctly handled and is unaffected._
 

package/README.ja.md

@@ -283,11 +283,11 @@ CI は `.github/workflows/npm-publish.yml` で公開します — GitHub リリ
 ## 変更履歴 (Changelog)
 
 <!-- CHANGELOG:START (auto-generated from CHANGELOG.md — run `bun run changelog:sync`) -->
+- **[0.8.34]** (2026-07-21) — Provider/model-catalog sync against gjc's actual current source (a shallow clone of the real, public `Yeachan-Heo/gajae-code` repo at v0.11.5 — direct `packages/ai/src/models.json` comparison, not inference from release notes). The system prompt's philosophy, the `/route` command, and the `tencent` provider were explicitly out of scope and are unchanged.
+- **[0.8.33]** (2026-07-20) — Real terminal-resize testing (repeated tmux resize-window + keystroke reproductions, byte-for-byte ANSI replay outside jeo/tmux to isolate root causes deterministically) uncovered and closed a progressive screen-corruption bug: resizing down to a narrow terminal (e.g. a ~20-column tmux pane) produced a growing stack of duplicate status-bar lines on every subsequent keystroke, and a separate, independently-reproducible scroll bug in the mid-turn live renderer.
 - **[0.8.32]** (2026-07-20) — gjc parity: TUI display-width/wrapping is now grapheme-cluster-aware for emoji sequences (VS16 presentation, skin-tone modifiers, keycaps, ZWJ-joined emoji like family/profession glyphs), fixing box-border and wrap misalignment ("깨짐") that the old per-code-point width summation produced whenever such a sequence appeared in a message — Korean/CJK-only text was already correctly handled and is unaffected.
 - **[0.8.31]** (2026-07-20) — `web_search` no longer silently degrades every non-Anthropic model to keyless DuckDuckGo scraping — OpenAI and Gemini sessions now get their own native, hosted search tool (matching the same active-model-gated, credential-required design Anthropic already had), with DuckDuckGo remaining the always-on terminal fallback for everyone else.
 - **[0.8.30]** (2026-07-20) — `bun run build` (the host-only dev binary build) has been silently broken for every user who ran it — a drift between the plain `package.json` `build` script and the actual working release-binary build command (`scripts/ci-release-build-binaries.ts`), which had already documented and fixed the exact same crash.
-- **[0.8.29]** (2026-07-20) — gajae-code (gjc) v0.10.2→v0.11.4 gap analysis: most of that range is gjc-specific architecture jeo intentionally does not replicate (SDK broker/session-index recovery, Gajae Pet, coordinator-mcp, Telegram/Discord rich-delivery internals, browser tab workers, worktree-subcommand removal, RPC durable model selection) — one genuinely applicable TUI bug was found and closed; the CLI empty-non-TTY-stdin fix (gjc #2586) was investigated and found already-correct by construction (jeo's one-shot gate forces true whenever stdin isn't a TTY, regardless of args, so the hang gjc fixed cannot occur here).
-- **[0.8.28]** (2026-07-16) — Improved the welcome UI's right-side table to display dynamic sections (What's New, Flow keys, Project pulse, and Session trail) matching gajae-code (gjc) >= 0.8.0 features.
 
 See [CHANGELOG.md](CHANGELOG.md) for the full history.
 <!-- CHANGELOG:END -->

package/README.ko.md

@@ -284,11 +284,11 @@ CI는 `.github/workflows/npm-publish.yml`로 배포합니다 — GitHub 릴리
 ## 변경 이력 (Changelog)
 
 <!-- CHANGELOG:START (auto-generated from CHANGELOG.md — run `bun run changelog:sync`) -->
+- **[0.8.34]** (2026-07-21) — Provider/model-catalog sync against gjc's actual current source (a shallow clone of the real, public `Yeachan-Heo/gajae-code` repo at v0.11.5 — direct `packages/ai/src/models.json` comparison, not inference from release notes). The system prompt's philosophy, the `/route` command, and the `tencent` provider were explicitly out of scope and are unchanged.
+- **[0.8.33]** (2026-07-20) — Real terminal-resize testing (repeated tmux resize-window + keystroke reproductions, byte-for-byte ANSI replay outside jeo/tmux to isolate root causes deterministically) uncovered and closed a progressive screen-corruption bug: resizing down to a narrow terminal (e.g. a ~20-column tmux pane) produced a growing stack of duplicate status-bar lines on every subsequent keystroke, and a separate, independently-reproducible scroll bug in the mid-turn live renderer.
 - **[0.8.32]** (2026-07-20) — gjc parity: TUI display-width/wrapping is now grapheme-cluster-aware for emoji sequences (VS16 presentation, skin-tone modifiers, keycaps, ZWJ-joined emoji like family/profession glyphs), fixing box-border and wrap misalignment ("깨짐") that the old per-code-point width summation produced whenever such a sequence appeared in a message — Korean/CJK-only text was already correctly handled and is unaffected.
 - **[0.8.31]** (2026-07-20) — `web_search` no longer silently degrades every non-Anthropic model to keyless DuckDuckGo scraping — OpenAI and Gemini sessions now get their own native, hosted search tool (matching the same active-model-gated, credential-required design Anthropic already had), with DuckDuckGo remaining the always-on terminal fallback for everyone else.
 - **[0.8.30]** (2026-07-20) — `bun run build` (the host-only dev binary build) has been silently broken for every user who ran it — a drift between the plain `package.json` `build` script and the actual working release-binary build command (`scripts/ci-release-build-binaries.ts`), which had already documented and fixed the exact same crash.
-- **[0.8.29]** (2026-07-20) — gajae-code (gjc) v0.10.2→v0.11.4 gap analysis: most of that range is gjc-specific architecture jeo intentionally does not replicate (SDK broker/session-index recovery, Gajae Pet, coordinator-mcp, Telegram/Discord rich-delivery internals, browser tab workers, worktree-subcommand removal, RPC durable model selection) — one genuinely applicable TUI bug was found and closed; the CLI empty-non-TTY-stdin fix (gjc #2586) was investigated and found already-correct by construction (jeo's one-shot gate forces true whenever stdin isn't a TTY, regardless of args, so the hang gjc fixed cannot occur here).
-- **[0.8.28]** (2026-07-16) — Improved the welcome UI's right-side table to display dynamic sections (What's New, Flow keys, Project pulse, and Session trail) matching gajae-code (gjc) >= 0.8.0 features.
 
 See [CHANGELOG.md](CHANGELOG.md) for the full history.
 <!-- CHANGELOG:END -->

package/README.md

@@ -291,11 +291,11 @@ Huge thanks to [gajae-code](https://github.com/Yeachan-Heo/gajae-code) for the i
 ## Changelog
 
 <!-- CHANGELOG:START (auto-generated from CHANGELOG.md — run `bun run changelog:sync`) -->
+- **[0.8.34]** (2026-07-21) — Provider/model-catalog sync against gjc's actual current source (a shallow clone of the real, public `Yeachan-Heo/gajae-code` repo at v0.11.5 — direct `packages/ai/src/models.json` comparison, not inference from release notes). The system prompt's philosophy, the `/route` command, and the `tencent` provider were explicitly out of scope and are unchanged.
+- **[0.8.33]** (2026-07-20) — Real terminal-resize testing (repeated tmux resize-window + keystroke reproductions, byte-for-byte ANSI replay outside jeo/tmux to isolate root causes deterministically) uncovered and closed a progressive screen-corruption bug: resizing down to a narrow terminal (e.g. a ~20-column tmux pane) produced a growing stack of duplicate status-bar lines on every subsequent keystroke, and a separate, independently-reproducible scroll bug in the mid-turn live renderer.
 - **[0.8.32]** (2026-07-20) — gjc parity: TUI display-width/wrapping is now grapheme-cluster-aware for emoji sequences (VS16 presentation, skin-tone modifiers, keycaps, ZWJ-joined emoji like family/profession glyphs), fixing box-border and wrap misalignment ("깨짐") that the old per-code-point width summation produced whenever such a sequence appeared in a message — Korean/CJK-only text was already correctly handled and is unaffected.
 - **[0.8.31]** (2026-07-20) — `web_search` no longer silently degrades every non-Anthropic model to keyless DuckDuckGo scraping — OpenAI and Gemini sessions now get their own native, hosted search tool (matching the same active-model-gated, credential-required design Anthropic already had), with DuckDuckGo remaining the always-on terminal fallback for everyone else.
 - **[0.8.30]** (2026-07-20) — `bun run build` (the host-only dev binary build) has been silently broken for every user who ran it — a drift between the plain `package.json` `build` script and the actual working release-binary build command (`scripts/ci-release-build-binaries.ts`), which had already documented and fixed the exact same crash.
-- **[0.8.29]** (2026-07-20) — gajae-code (gjc) v0.10.2→v0.11.4 gap analysis: most of that range is gjc-specific architecture jeo intentionally does not replicate (SDK broker/session-index recovery, Gajae Pet, coordinator-mcp, Telegram/Discord rich-delivery internals, browser tab workers, worktree-subcommand removal, RPC durable model selection) — one genuinely applicable TUI bug was found and closed; the CLI empty-non-TTY-stdin fix (gjc #2586) was investigated and found already-correct by construction (jeo's one-shot gate forces true whenever stdin isn't a TTY, regardless of args, so the hang gjc fixed cannot occur here).
-- **[0.8.28]** (2026-07-16) — Improved the welcome UI's right-side table to display dynamic sections (What's New, Flow keys, Project pulse, and Session trail) matching gajae-code (gjc) >= 0.8.0 features.
 
 See [CHANGELOG.md](CHANGELOG.md) for the full history.
 <!-- CHANGELOG:END -->

package/README.zh.md

@@ -283,11 +283,11 @@ CI 通过 `.github/workflows/npm-publish.yml` 发布 — GitHub 发布 release
 ## 更新日志 (Changelog)
 
 <!-- CHANGELOG:START (auto-generated from CHANGELOG.md — run `bun run changelog:sync`) -->
+- **[0.8.34]** (2026-07-21) — Provider/model-catalog sync against gjc's actual current source (a shallow clone of the real, public `Yeachan-Heo/gajae-code` repo at v0.11.5 — direct `packages/ai/src/models.json` comparison, not inference from release notes). The system prompt's philosophy, the `/route` command, and the `tencent` provider were explicitly out of scope and are unchanged.
+- **[0.8.33]** (2026-07-20) — Real terminal-resize testing (repeated tmux resize-window + keystroke reproductions, byte-for-byte ANSI replay outside jeo/tmux to isolate root causes deterministically) uncovered and closed a progressive screen-corruption bug: resizing down to a narrow terminal (e.g. a ~20-column tmux pane) produced a growing stack of duplicate status-bar lines on every subsequent keystroke, and a separate, independently-reproducible scroll bug in the mid-turn live renderer.
 - **[0.8.32]** (2026-07-20) — gjc parity: TUI display-width/wrapping is now grapheme-cluster-aware for emoji sequences (VS16 presentation, skin-tone modifiers, keycaps, ZWJ-joined emoji like family/profession glyphs), fixing box-border and wrap misalignment ("깨짐") that the old per-code-point width summation produced whenever such a sequence appeared in a message — Korean/CJK-only text was already correctly handled and is unaffected.
 - **[0.8.31]** (2026-07-20) — `web_search` no longer silently degrades every non-Anthropic model to keyless DuckDuckGo scraping — OpenAI and Gemini sessions now get their own native, hosted search tool (matching the same active-model-gated, credential-required design Anthropic already had), with DuckDuckGo remaining the always-on terminal fallback for everyone else.
 - **[0.8.30]** (2026-07-20) — `bun run build` (the host-only dev binary build) has been silently broken for every user who ran it — a drift between the plain `package.json` `build` script and the actual working release-binary build command (`scripts/ci-release-build-binaries.ts`), which had already documented and fixed the exact same crash.
-- **[0.8.29]** (2026-07-20) — gajae-code (gjc) v0.10.2→v0.11.4 gap analysis: most of that range is gjc-specific architecture jeo intentionally does not replicate (SDK broker/session-index recovery, Gajae Pet, coordinator-mcp, Telegram/Discord rich-delivery internals, browser tab workers, worktree-subcommand removal, RPC durable model selection) — one genuinely applicable TUI bug was found and closed; the CLI empty-non-TTY-stdin fix (gjc #2586) was investigated and found already-correct by construction (jeo's one-shot gate forces true whenever stdin isn't a TTY, regardless of args, so the hang gjc fixed cannot occur here).
-- **[0.8.28]** (2026-07-16) — Improved the welcome UI's right-side table to display dynamic sections (What's New, Flow keys, Project pulse, and Session trail) matching gajae-code (gjc) >= 0.8.0 features.
 
 See [CHANGELOG.md](CHANGELOG.md) for the full history.
 <!-- CHANGELOG:END -->

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "jeo-code",
-  "version": "0.8.32",
+  "version": "0.8.34",
   "description": "Clean, highly optimized AI coding agent using spec-first loop",
   "type": "module",
   "main": "src/cli.ts",

package/src/ai/providers/openai-compatible-catalog.ts

@@ -59,6 +59,13 @@ export const OPENAI_COMPAT_PROVIDERS: readonly OpenAICompatProviderDef[] = [
   { name: "zenmux", label: "ZenMux", baseUrl: "https://zenmux.ai/api/v1", apiKeyEnv: "ZENMUX_API_KEY", defaultModel: "zenmux/anthropic/claude-opus-4.6" },
   { name: "qianfan", label: "Qianfan", baseUrl: "https://qianfan.baidubce.com/v2", apiKeyEnv: "QIANFAN_API_KEY", defaultModel: "qianfan/deepseek-v3.2" },
   { name: "xiaomi", label: "Xiaomi", baseUrl: "https://api.xiaomimimo.com/v1", apiKeyEnv: "XIAOMI_API_KEY", defaultModel: "xiaomi/mimo-v2-flash" },
+  { name: "deepinfra", label: "DeepInfra", baseUrl: "https://api.deepinfra.com/v1/openai", apiKeyEnv: "DEEPINFRA_API_KEY", defaultModel: "deepinfra/deepseek-ai/DeepSeek-V3.2", knownModels: ["deepseek-ai/DeepSeek-V3.2", "deepseek-ai/DeepSeek-R1-0528", "meta-llama/Llama-3.3-70B-Instruct-Turbo", "moonshotai/Kimi-K2.6", "Qwen/Qwen3-Coder-480B-A35B-Instruct-Turbo", "openai/gpt-oss-120b"] },
+  // Self-hosted OpenAI-compatible proxy (LiteLLM) — the model catalog behind it is
+  // entirely user-configured (whatever the user's own litellm config.yaml routes to),
+  // so unlike every other row here there is no globally-fixed `knownModels` list to
+  // offer; `baseUrl` is LiteLLM's own documented local-proxy default and is expected
+  // to be overridden via `jeo provider add litellm --base-url <your-proxy-url>`.
+  { name: "litellm", label: "LiteLLM", baseUrl: "http://localhost:4000/v1", apiKeyEnv: "LITELLM_API_KEY", defaultModel: "litellm/gpt-4o-mini" },
   { name: "xiaomi-token-plan-ams", label: "Xiaomi Token Plan (Europe)", baseUrl: "https://token-plan-ams.xiaomimimo.com/v1", apiKeyEnv: "XIAOMI_TOKEN_PLAN_AMS_API_KEY", defaultModel: "xiaomi-token-plan-ams/mimo-v2.5", subscription: true },
   { name: "xiaomi-token-plan-cn", label: "Xiaomi Token Plan (China)", baseUrl: "https://token-plan-cn.xiaomimimo.com/v1", apiKeyEnv: "XIAOMI_TOKEN_PLAN_CN_API_KEY", defaultModel: "xiaomi-token-plan-cn/mimo-v2.5", subscription: true },
   { name: "xiaomi-token-plan-sgp", label: "Xiaomi Token Plan (Singapore)", baseUrl: "https://token-plan-sgp.xiaomimimo.com/v1", apiKeyEnv: "XIAOMI_TOKEN_PLAN_SGP_API_KEY", defaultModel: "xiaomi-token-plan-sgp/mimo-v2.5", subscription: true },

package/src/ai/types.ts

@@ -1,6 +1,6 @@
 import type { Credential } from "../auth";
 
-export type ProviderName = "anthropic" | "openai" | "gemini" | "antigravity" | "ollama" | "lmstudio" | "xai" | "kimi" | "groq" | "deepseek" | "mistral" | "openrouter" | "together" | "cerebras" | "fireworks" | "nvidia" | "alibaba-coding-plan" | "huggingface" | "nanogpt" | "qwen-portal" | "synthetic" | "venice" | "zenmux" | "qianfan" | "xiaomi" | "xiaomi-token-plan-ams" | "xiaomi-token-plan-cn" | "xiaomi-token-plan-sgp" | "minimax-code" | "minimax-code-cn" | "zai" | "minimax" | "tencent";
+export type ProviderName = "anthropic" | "openai" | "gemini" | "antigravity" | "ollama" | "lmstudio" | "xai" | "kimi" | "groq" | "deepseek" | "mistral" | "openrouter" | "together" | "cerebras" | "fireworks" | "nvidia" | "alibaba-coding-plan" | "huggingface" | "nanogpt" | "qwen-portal" | "synthetic" | "venice" | "zenmux" | "qianfan" | "xiaomi" | "xiaomi-token-plan-ams" | "xiaomi-token-plan-cn" | "xiaomi-token-plan-sgp" | "minimax-code" | "minimax-code-cn" | "zai" | "minimax" | "tencent" | "deepinfra" | "litellm";
 
 
 /** An image attached to a (user) message — base64 payload + IANA media type. */

package/src/auth/storage.ts

@@ -8,12 +8,12 @@ import { jeoEnv } from "../util/env";
 /** Providers with an interactive OAuth login + refresh flow. */
 export type OAuthProvider = "anthropic" | "openai" | "gemini" | "antigravity" | "kimi";
 /** Every provider jeo resolves a credential for: OAuth-capable ∪ API-key-only. */
-export type AuthProvider = OAuthProvider | "xai" | "groq" | "deepseek" | "mistral" | "openrouter" | "together" | "cerebras" | "fireworks" | "nvidia" | "alibaba-coding-plan" | "huggingface" | "nanogpt" | "qwen-portal" | "synthetic" | "venice" | "zenmux" | "qianfan" | "xiaomi" | "xiaomi-token-plan-ams" | "xiaomi-token-plan-cn" | "xiaomi-token-plan-sgp" | "minimax-code" | "minimax-code-cn" | "zai" | "minimax" | "tencent";
+export type AuthProvider = OAuthProvider | "xai" | "groq" | "deepseek" | "mistral" | "openrouter" | "together" | "cerebras" | "fireworks" | "nvidia" | "alibaba-coding-plan" | "huggingface" | "nanogpt" | "qwen-portal" | "synthetic" | "venice" | "zenmux" | "qianfan" | "xiaomi" | "xiaomi-token-plan-ams" | "xiaomi-token-plan-cn" | "xiaomi-token-plan-sgp" | "minimax-code" | "minimax-code-cn" | "zai" | "minimax" | "tencent" | "deepinfra" | "litellm";
 
 
 export const OAUTH_PROVIDERS: readonly OAuthProvider[] = ["anthropic", "openai", "gemini", "antigravity", "kimi"];
 /** API-key-only providers (no OAuth flow) — resolved from config.providers / `<NAME>_API_KEY`. */
-export const API_KEY_ONLY_PROVIDERS: readonly AuthProvider[] = ["xai", "groq", "deepseek", "mistral", "openrouter", "together", "cerebras", "fireworks", "nvidia", "alibaba-coding-plan", "huggingface", "nanogpt", "qwen-portal", "synthetic", "venice", "zenmux", "qianfan", "xiaomi", "xiaomi-token-plan-ams", "xiaomi-token-plan-cn", "xiaomi-token-plan-sgp", "minimax-code", "minimax-code-cn", "zai", "minimax", "tencent"];
+export const API_KEY_ONLY_PROVIDERS: readonly AuthProvider[] = ["xai", "groq", "deepseek", "mistral", "openrouter", "together", "cerebras", "fireworks", "nvidia", "alibaba-coding-plan", "huggingface", "nanogpt", "qwen-portal", "synthetic", "venice", "zenmux", "qianfan", "xiaomi", "xiaomi-token-plan-ams", "xiaomi-token-plan-cn", "xiaomi-token-plan-sgp", "minimax-code", "minimax-code-cn", "zai", "minimax", "tencent", "deepinfra", "litellm"];
 
 /** Narrow an AuthProvider to the OAuth-capable subset (xai has no OAuth flow). */
 export function isOAuthProvider(p: AuthProvider): p is OAuthProvider {

package/src/commands/launch.ts

@@ -764,7 +764,7 @@ export async function runLaunchCommand(args: string[]): Promise<void> {
    *  mid-turn TUI binding): full last reply + full last tool output. */
   const composeDetailLines = (): string[] => {
     if (!lastReply && !lastToolDetail) return [];
-    const sep = "─".repeat(Math.min(48, Math.max(20, (process.stdout.columns ?? 80) - 1)));
+    const sep = "─".repeat(Math.min(48, Math.max(1, (process.stdout.columns ?? 80) - 1)));
     const toolDetail = lastToolDetail
       ? [sep, `detail · full last tool output (${lastToolDetail.tool})`, sep, ...lastToolDetail.output.split("\n").slice(0, 2000).map(sanitizeForTerminal)]
       : [];
@@ -816,7 +816,7 @@ export async function runLaunchCommand(args: string[]): Promise<void> {
             // exact prior conversation instead of a bare message count.
             if (messages.length > 0) {
               resumedWithHistory = true;
-              const sep = "─".repeat(Math.min(48, Math.max(20, (process.stdout.columns ?? 80) - 1)));
+              const sep = "─".repeat(Math.min(48, Math.max(1, (process.stdout.columns ?? 80) - 1)));
               console.log([sep, ...formatTranscript(history, { maxTurns: 6, color: true, unicode: true }), sep].join("\n"));
             }
 
@@ -880,7 +880,7 @@ export async function runLaunchCommand(args: string[]): Promise<void> {
           // screen instead of just a bare message count.
           if (messages.length > 0) {
             resumedWithHistory = true;
-            const sep = "─".repeat(Math.min(48, Math.max(20, (process.stdout.columns ?? 80) - 1)));
+            const sep = "─".repeat(Math.min(48, Math.max(1, (process.stdout.columns ?? 80) - 1)));
             console.log([sep, ...formatTranscript(history, { maxTurns: 6, color: true, unicode: true }), sep].join("\n"));
           }
 
@@ -2209,7 +2209,7 @@ export async function runLaunchCommand(args: string[]): Promise<void> {
   // so it never repeats. Screen-safe: prints BEFORE the prompt is armed.
   try {
     const whatsNew = await consumeLaunchWhatsNew({
-      cols: Math.min(100, Math.max(40, (process.stdout.columns ?? 80) - 2)),
+      cols: Math.min(100, Math.max(1, (process.stdout.columns ?? 80) - 2)),
       unicode: supportsUnicode(),
       color: welcomeTheme.color,
     });
@@ -2224,7 +2224,7 @@ export async function runLaunchCommand(args: string[]): Promise<void> {
     {
       const card = formatForgeBox(
         { title: "[skill]", lines: skillInvocationCard(skill, intent) },
-        { width: scaleForgeWidth(Math.min(100, Math.max(40, (process.stdout.columns ?? 80) - 2))), unicode: supportsUnicode(), paint: accentPaint(uiTheme), paintShadow: accentShadowPaint(uiTheme), color: uiTheme.color },
+        { width: scaleForgeWidth(Math.min(100, Math.max(1, (process.stdout.columns ?? 80) - 2))), unicode: supportsUnicode(), paint: accentPaint(uiTheme), paintShadow: accentShadowPaint(uiTheme), color: uiTheme.color },
       );
       logLines(card);
     }
@@ -2729,7 +2729,7 @@ export async function runLaunchCommand(args: string[]): Promise<void> {
   let lastFooterKey = "";
   const logLines = (lines: string | string[]) => {
     const arr = Array.isArray(lines) ? lines : [lines];
-    const cols = Math.max(20, (process.stdout.columns ?? 80) - 1);
+    const cols = Math.max(1, (process.stdout.columns ?? 80) - 1);
     for (const line of arr) {
       console.log(truncateAnsi(line, cols));
     }
@@ -2934,7 +2934,7 @@ export async function runLaunchCommand(args: string[]): Promise<void> {
     });
   };
   const previewLines = (line: string, selected = -1): string[] => {
-    const cols = Math.max(24, (process.stdout.columns ?? 80) - 1);
+    const cols = Math.max(1, (process.stdout.columns ?? 80) - 1);
     // Caret offset comes from readline's live cursor when it matches the rendered
     // line (arrow keys/Home/End move it); otherwise (history nav mismatch) caret
     // sits at the end of the text.
@@ -2983,7 +2983,7 @@ export async function runLaunchCommand(args: string[]): Promise<void> {
   // with ↑/↓ counters. Recomputes the scroll bound/page size each paint so scroll
   // keys can clamp correctly. Mirrors the mid-turn LaunchTui panel.
   const historyPreviewLines = (detail: string[]): string[] => {
-    const cols = Math.max(24, (process.stdout.columns ?? 80) - 1);
+    const cols = Math.max(1, (process.stdout.columns ?? 80) - 1);
     const physical = detail.flatMap(line => line.split("\n")).map(line => truncateAnsi(line, cols));
     const bodyLimit = Math.max(1, footerRows - 2); // status bar + title rows
     const scrollable = physical.length > bodyLimit;

package/src/tui/components/input-box.ts

@@ -107,7 +107,13 @@ function normalizeHighlights(
  * can park the REAL terminal cursor right after `>` — moving with the arrow keys.
  */
 export function renderInputFrame(line: string, opts: InputBoxOptions = {}): InputFrame {
-  const cols = Math.max(24, Math.trunc(opts.cols ?? 80));
+  // Never clamp UP past what the caller reports as the real terminal width — a
+  // fixed floor here (previously 24) silently overflowed any narrower terminal
+  // (e.g. a resize down to 20 cols still drew a 24-col box), which real terminals
+  // then hard-wrap/split at the boundary (the exact box-border corruption a real
+  // resize-down reproduces live). Degrading to a very small, ugly-but-correctly-
+  // sized box is strictly better than one that overflows and gets torn.
+  const cols = Math.max(1, Math.trunc(opts.cols ?? 80));
   const useColor = opts.color !== false;
   const placeholder = opts.placeholder ?? "Type your message...";
   const bodyWidth = Math.max(1, cols - 4);

package/src/tui/components/status.ts

@@ -66,7 +66,14 @@ function shortenCwd(cwd: string, maxLen: number, unicode: boolean): string {
 export function renderStatusBar(d: StatusBarData): string {
   const unicode = d.unicode !== false;
   const useColor = d.color !== false;
-  const cols = Math.max(24, Math.trunc(d.cols));
+  // Never clamp UP past the caller's real terminal width — a fixed floor here
+  // (previously 24) silently overflowed any narrower terminal (e.g. a resize down
+  // to 20 cols still emitted a 24-col status bar), which a real terminal then
+  // hard-wraps at the boundary and desyncs the idle-footer's own row bookkeeping
+  // on every subsequent repaint (reproduced live: a growing stack of duplicate
+  // status-bar lines on each keystroke after a resize-down, root-caused to this
+  // exact overflow via a byte-for-byte ANSI replay outside jeo/tmux entirely).
+  const cols = Math.max(1, Math.trunc(d.cols));
   const sep = " / ";
 
   // Right side first — it has priority and fixed width.
@@ -97,7 +104,7 @@ export function renderStatusBar(d: StatusBarData): string {
     bits.push(`${unicode ? "⑂" : "y"} ${d.branch}${dirty}`);
   }
   // Budget the cwd into whatever width remains (right stats + 2-col gap reserved).
-  const leftBudget = Math.max(8, cols - rightWidth - (rightWidth > 0 ? 2 : 0));
+  const leftBudget = Math.max(1, cols - rightWidth - (rightWidth > 0 ? 2 : 0));
   if (d.cwd) {
     const used = visibleWidth(` ${bits.join(sep)}${sep}${unicode ? "▸" : ">"}  `);
     const room = leftBudget - used - 1;

package/src/tui/components/welcome.ts

@@ -185,7 +185,17 @@ export function renderWelcome(d: WelcomeData): string[] {
   const useColor = d.color !== false;
 
   if (cols < 30) {
-    return [ `jeo v${d.version} · ${d.model}` ];
+    // Narrow-terminal one-line fallback — MUST itself be width-capped: an unbounded
+    // `jeo v{version} · {model}` overflows badly with a realistic provider-qualified
+    // model id (e.g. "antigravity/claude-sonnet-4-6 (antigravity)" overflows a 10-col
+    // terminal by 47 columns), which is exactly the box-border/wrap misalignment a
+    // real terminal-resize-down reproduces live. `cols - 1` (not `cols`) — same
+    // "leave the last column free" convention as the boxed banner below: a line
+    // truncated to EXACTLY `cols` is ambiguous to real terminals (a full-width row
+    // followed by this line's own trailing "\n" can double-advance a row via the
+    // pending-autowrap/explicit-LF ambiguity), which is a second, independent
+    // corruption vector reproduced live — not just a cosmetic nicety.
+    return [truncate(`jeo v${d.version} · ${d.model}`, Math.max(0, cols - 1))];
   }
 
   // The banner fills the full terminal width (gjc forge: flush with the input box and

package/src/tui/renderer.ts

@@ -49,7 +49,29 @@ export class Renderer {
     const currentRows = size().rows;
     if ((this.prevCols !== undefined && this.prevCols !== currentCols) ||
         (this.prevRows !== undefined && this.prevRows !== currentRows)) {
-      this.clear();
+      if (this.reserve) {
+        // Inline/reserve mode: a resize needs a full repaint, but NOT an immediate
+        // clear() — clear() zeroes coverRows/prev to 0 as an independent write, which
+        // then fools the reserve block below (right after this) into believing NOTHING
+        // currently occupies the screen (occupied=0) — even though the old frame's rows
+        // are still physically there. With occupied wrongly 0, `next.length > occupied`
+        // spuriously trips even when the new (post-resize) frame is SHORTER than the
+        // old one, inserting real "\n" characters that ACTUALLY SCROLL the terminal —
+        // corrupting whatever sits above the live frame and permanently desyncing this
+        // renderer's row bookkeeping from reality (every later diff then paints at the
+        // wrong absolute row, producing progressively worse duplicate/torn content on
+        // each subsequent tick — reproduced deterministically with a standalone
+        // Renderer harness before this fix landed). reset() is the same self-heal path
+        // the periodic full-resync already uses: it remembers the old occupied rows via
+        // coverRows (no output), so the diff loop below EL-clears exactly the excess
+        // rows in place — no scroll, no desync.
+        this.reset();
+      } else {
+        // Non-reserve (alt-screen/pipe) mode has no reserve block to fool (gated on
+        // `this.reserve`), and the underlying screen buffer's own dimensions just
+        // changed — keep the defensive full ED clear.
+        this.clear();
+      }
     }
     this.prevCols = currentCols;
     this.prevRows = currentRows;