@nanobpm/nano-coder 0.35.0 → 0.37.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (2) hide show
  1. package/README.md +94 -7
  2. package/package.json +5 -5
package/README.md CHANGED
@@ -345,7 +345,7 @@ model where it may write.
345
345
  Typing `/` at the prompt lists the commands under it, and each further character narrows the
346
346
  list. Tab completes the command, or the part all matches share. Esc hides the list. The
347
347
  list is built from the same table as `/help` (`src/commands.rs`). Commands with a known
348
- argument set (`/model`, `/mode`, `/verbosity`) get the same treatment for their first
348
+ argument set (`/model`, `/mode`, `/verbosity`, `/thinking`) get the same treatment for their first
349
349
  argument: a type-ahead list narrows as you type and Tab completes it.
350
350
 
351
351
  - `/help` - Show available commands
@@ -379,6 +379,7 @@ argument: a type-ahead list narrows as you type and Tab completes it.
379
379
  - **normal** - full tools; reaching a positive turn cap asks whether to keep going
380
380
  - **plan** - read-only: mutating tools (`bash`, `write_file`, `edit_file`) are gated, only analysis and output
381
381
  - **auto** - no turn cap; a `question` left unanswered for 15s is answered with "the user is away from the keyboard, make the best decision you can"
382
+ - `/thinking [level|default|off|reset]` - Show the thinking level in use and the levels the current model takes, or set one for this session (`reset` goes back to the configured level). See [Thinking](#thinking)
382
383
  - `/providers` - List providers, endpoints and whether their API key is available
383
384
  - `/session` - Show the session ID and log path
384
385
  - `/trajectory` - Show this session's trajectory turn by turn: user input, thinking, answers, tool calls and results, tokens and timings. Each message row is labelled with its `#N` session-log ID, the same ID `history_read` and smart-compaction summaries use (compaction and crash-recovered input rows have no `#N`, as they aren't cited that way). When it doesn't fit on the screen it opens in your pager (`$PAGER`, default `less`) — but only at an idle prompt with the frame renderer: invoked mid-turn (while a turn runs) or under the legacy renderer it prints inline instead. `/trajectory --json` or `/trajectory --markdown` prints an export instead; `nano-coder --trajectory SESSION_ID [--json|--markdown]` does the same for any saved session
@@ -508,7 +509,7 @@ Built-in presets:
508
509
  | `kimi` | openai | `https://api.moonshot.ai/v1` | `MOONSHOT_API_KEY` |
509
510
  | `mistral` | openai | `https://api.mistral.ai/v1` | `MISTRAL_API_KEY` |
510
511
  | `gemini` | openai | `https://generativelanguage.googleapis.com/v1beta/openai` | `GEMINI_API_KEY` |
511
- | `qwen` | openai | `https://dashscope-intl.aliyuncs.com/compatible-mode/v1` | `DASHSCOPE_API_KEY` |
512
+ | `qwen` | openai | Model Studio (default: Standard, Singapore — all plans/regions below) | `DASHSCOPE_API_KEY` / `BAILIAN_*_PLAN_API_KEY` |
512
513
  | `ollama` | openai | `http://localhost:11434/v1` | — |
513
514
  | `llamacpp` | openai | `http://localhost:8080/v1` | — |
514
515
  | `github-copilot` | github-copilot | from session token | `GITHUB_COPILOT_OAUTH_TOKEN` or `--login` (unofficial, see below) |
@@ -546,11 +547,40 @@ timeout_secs = 300 # idle timeout: max silence between stre
546
547
  max_retries = 3
547
548
  ```
548
549
 
549
- `qwen` is Qwen Cloud (Alibaba Cloud Model Studio), e.g. `--model qwen/qwen3.8-max`. The
550
- preset uses the Singapore endpoint. API keys are bound to a region, so for another region
551
- or your workspace domain override `base_url`, e.g.
552
- `https://dashscope-us.aliyuncs.com/compatible-mode/v1` or
553
- `https://<WorkspaceId>.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`.
550
+ `qwen` is Qwen Cloud (Alibaba Cloud Model Studio), e.g. `--model qwen/qwen3.8-max`.
551
+ Model Studio serves the same models through three plans, each with its own hostnames and
552
+ API-key variable. All of them speak OpenAI Chat Completions, so each is just a `base_url`
553
+ + `api_key_env`:
554
+
555
+ | Plan | Region | Base URL | API key env |
556
+ |---|---|---|---|
557
+ | Standard API key | Singapore (International) | `https://dashscope-intl.aliyuncs.com/compatible-mode/v1` | `DASHSCOPE_API_KEY` |
558
+ | Standard API key | China (Beijing) | `https://dashscope.aliyuncs.com/compatible-mode/v1` | `DASHSCOPE_API_KEY` |
559
+ | Standard API key | US (Virginia) | `https://dashscope-us.aliyuncs.com/compatible-mode/v1` | `DASHSCOPE_API_KEY` |
560
+ | Standard API key | China (Hong Kong) | `https://cn-hongkong.dashscope.aliyuncs.com/compatible-mode/v1` | `DASHSCOPE_API_KEY` |
561
+ | Token Plan | Singapore (International) | `https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` | `BAILIAN_TOKEN_PLAN_API_KEY` |
562
+ | Token Plan | China (Beijing) | `https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` | `BAILIAN_TOKEN_PLAN_API_KEY` |
563
+ | Coding Plan | Singapore (International) | `https://coding-intl.dashscope.aliyuncs.com/v1` | `BAILIAN_CODING_PLAN_API_KEY` |
564
+ | Coding Plan | China (Beijing) | `https://coding.dashscope.aliyuncs.com/v1` | `BAILIAN_CODING_PLAN_API_KEY` |
565
+
566
+ The `qwen` preset defaults to **Standard API key, Singapore** (the first row). In
567
+ `/settings` → *Add or edit a provider* → `qwen` these eight endpoints are offered as a
568
+ *Qwen / Model Studio endpoint* picker (plus a custom URL), so you can switch plan and
569
+ region without retyping a URL. Picking a plan also points the provider at that plan's
570
+ API-key variable — the preset's `DASHSCOPE_API_KEY` becomes `BAILIAN_TOKEN_PLAN_API_KEY`
571
+ for Token Plan, and `BAILIAN_CODING_PLAN_API_KEY` for Coding Plan. Set the variable (or a
572
+ key command / literal key) at the key prompt that follows. API keys are also bound to a
573
+ *region*, so a key issued for another region still needs the matching endpoint row.
574
+
575
+ To configure an endpoint in the config file directly, override `base_url` and
576
+ `api_key_env` (append `/compatible-mode/v1` to a workspace domain, e.g.
577
+ `https://<WorkspaceId>.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`):
578
+
579
+ ```toml
580
+ [providers.qwen] # Token Plan, China
581
+ base_url = "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1"
582
+ api_key_env = "BAILIAN_TOKEN_PLAN_API_KEY"
583
+ ```
554
584
 
555
585
  `kimi` is the Kimi API from platform.kimi.ai, e.g. `--model kimi/kimi-k3` or
556
586
  `kimi/kimi-k2.7-code`. The preset drops `temperature` (K3 fixes it) and sets
@@ -582,6 +612,63 @@ top-level `temperature` just doesn't apply to them. Anthropic accepts 0 to 1, so
582
612
  value is sent as 1, with a warning. `/context` shows the temperature in use and where it
583
613
  comes from, and `/settings` edits it for the current model, its provider, or all models.
584
614
 
615
+ ### Thinking
616
+
617
+ `thinking` sets how much the model reasons before it answers: `"default"` sends nothing
618
+ (the model decides), `"off"` turns thinking off, and a level such as `"low"`, `"medium"`,
619
+ `"high"`, `"xhigh"` or `"max"` asks for that much. Like `temperature`, it can be set for all
620
+ models, one provider or one model, and `/thinking LEVEL` overrides them for the session:
621
+
622
+ ```toml
623
+ thinking = "medium" # all models that support it
624
+
625
+ [providers.anthropic.models."claude-opus-4-7"]
626
+ thinking = "xhigh" # this model only
627
+ ```
628
+
629
+ A model only gets levels it supports. They come from, in order: `thinking_levels` on the
630
+ model or provider; what the endpoint reports for the model; and a built-in table
631
+ of Claude (3.7 Sonnet and later) and OpenAI reasoning models (GPT-5 and later, o1/o3/o4).
632
+ Endpoints that report levels:
633
+
634
+ - **GitHub Copilot:** `/models` lists each model's levels, read with the same request as its
635
+ context window.
636
+ - **Ollama:** a model with the `thinking` capability (`/api/show`) gets `off`, `low`,
637
+ `medium` and `high`. Ollama refuses a level for a model without it, so none is sent then.
638
+ - **llama.cpp:** the loaded model's chat template (`/props`) decides. A template with an
639
+ `enable_thinking` switch (Qwen 3 and the like) gets `off` and `on`, and any named level
640
+ turns thinking on; a template that takes `reasoning_effort` (gpt-oss) gets `low`, `medium`
641
+ and `high`.
642
+
643
+ Ollama and llama.cpp are recognized however the provider is named, the same way as for the
644
+ context window. For other models, list the levels yourself:
645
+
646
+ ```toml
647
+ [providers.together.models."deepseek-r1"]
648
+ thinking_levels = ["low", "medium", "high"] # add "off" if thinking can be turned off
649
+ ```
650
+
651
+ A level the model lacks is moved to the nearest one it has (the highest below it, else the
652
+ lowest), and a model with no known levels is sent nothing. A setting for the provider, the
653
+ model or the session warns when it is adjusted or ignored; the top-level `thinking` doesn't
654
+ warn, since it applies to every model.
655
+
656
+ How the level is sent depends on the API: `reasoning_effort` (Chat Completions, Ollama
657
+ included; Ollama ignores its native `think` flag on `/v1`), `chat_template_kwargs` for
658
+ llama.cpp (`enable_thinking`, or `reasoning_effort` for templates that take a level; it
659
+ ignores the top-level `reasoning_effort`),
660
+ `reasoning.effort` (Responses), and for Anthropic Messages adaptive thinking with
661
+ `output_config.effort`, or on Claude 3.7 to 4.5 a fixed `budget_tokens` (1024 for minimal, then
662
+ 2048, 8192, 16384, 32768, 65536 up to max). The budget has to stay below `max_tokens`, so raise
663
+ `max_tokens` to use a large one. Anthropic takes no custom temperature while thinking, so
664
+ none is sent then. A matching key in the provider's `extra_body` (`reasoning_effort`, `reasoning`,
665
+ `think`, `chat_template_kwargs`, `thinking`, `output_config`) is sent instead, with a warning; the
666
+ configured level is then reported as overridden (not as sent), since the override's value is what
667
+ reaches the wire. The status bar shows
668
+ `think LEVEL` while a level is sent, `/context` shows it with where it comes from,
669
+ `/settings` sets it for the current model, its provider or all models (picking from the
670
+ levels the model supports), and the session log records it for each reply.
671
+
585
672
  Other per-provider fields: `replay_reasoning`, `max_tokens_param` (`max_tokens`, or `max_completion_tokens`
586
673
  which is the `openai` default), `retry_initial_backoff_ms`, `retry_max_backoff_ms` and
587
674
  `retryable_statuses`.
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@nanobpm/nano-coder",
3
- "version": "0.35.0",
3
+ "version": "0.37.0",
4
4
  "description": "A 6MB coding agent. Run a fleet on your laptop.",
5
5
  "license": "Apache-2.0",
6
6
  "repository": {
@@ -31,9 +31,9 @@
31
31
  "node": ">=18"
32
32
  },
33
33
  "optionalDependencies": {
34
- "@nanobpm/nano-coder-darwin-arm64": "0.35.0",
35
- "@nanobpm/nano-coder-darwin-x64": "0.35.0",
36
- "@nanobpm/nano-coder-linux-arm64": "0.35.0",
37
- "@nanobpm/nano-coder-linux-x64": "0.35.0"
34
+ "@nanobpm/nano-coder-darwin-arm64": "0.37.0",
35
+ "@nanobpm/nano-coder-darwin-x64": "0.37.0",
36
+ "@nanobpm/nano-coder-linux-arm64": "0.37.0",
37
+ "@nanobpm/nano-coder-linux-x64": "0.37.0"
38
38
  }
39
39
  }