foldrun 13.2.0 → 14.0.1

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.
package/docs/providers.md CHANGED
@@ -51,6 +51,14 @@ billed in credits at **foldrun's price list**
51
51
  `foldrun account model prices`. Anthropic's tiers are the usual Claude
52
52
  models. A model id written out in full runs only if the price list names
53
53
  it; anything not listed is refused before the call is made.
54
+ - **Your `effort:` on each provider.** The agent runtime speaks Claude's
55
+ API; each provider is sent what its own API documents, the rest left
56
+ out. Anthropic gets the request as written. DeepSeek takes every level
57
+ (`medium` and `xhigh` run as `high`). Kimi takes `low`, `high` and
58
+ `max`: `medium` and `xhigh` run as `high`, and it always thinks. OpenAI
59
+ (over its Responses API) and Gemini take `low`, `medium`, `high`;
60
+ `xhigh` and `max` run as `high`. No provider is sent `temperature` or
61
+ `top_p` — the reasoning models fix them.
54
62
  - **Where each provider processes the prompt** is its own: DeepSeek in
55
63
  the People's Republic of China (its privacy policy), Moonshot in
56
64
  Singapore, the others as their own terms say. Choose the provider
@@ -125,7 +133,7 @@ preset, say); the file warns if you try.
125
133
  | `ollama` | Ollama (local) | anthropic | `http://localhost:11434` | x-api-key | | v0.14+ (2026-01). Ignores `tool_choice` and `cache_control`; accepts but does not enforce thinking budgets; base64 images only. Any key value is accepted. |
126
134
  | `lmstudio` | LM Studio (local) | anthropic | `http://localhost:1234` | x-api-key | | |
127
135
  | `vllm` | vLLM (yours) | anthropic | `(per account — set base_url)` | bearer | | base_url is your server (default :8000). Start it with `--enable-auto-tool-choice` and a `--tool-call-parser` or no tool loop closes. |
128
- | `openai` | OpenAI | openai | `https://api.openai.com/v1` | bearer | yes | Chat Completions by default, which OpenAI keeps supporting indefinitely. Add `format: responses` for the Responses API ("recommended for all new projects"): PDFs cross as input files and `reasoning.effort` is native; still every turn carries the whole conversation. |
136
+ | `openai` | OpenAI | responses | `https://api.openai.com/v1` | bearer | yes | The Responses API by default since 3 Nov 2026: GPT-6 Astra and GPT-6.1 Sol call tools there alone, and GPT-6 Luna on Chat Completions only with reasoning off. PDFs cross as input files and `reasoning.effort` is native; every turn still carries the whole conversation (`store: false`). `format: openai` says Chat Completions, for an older model or a proxy that speaks only that. |
129
137
  | `gemini` | Google Gemini | openai | `https://generativelanguage.googleapis.com/v1beta/openai` | bearer | | Google's OpenAI-compatible route. Unknown parameters are ignored silently; reasoning cannot be switched off on the newest models. |
130
138
  | `xai` | xAI Grok | openai | `https://api.x.ai/v1` | bearer | | Sends `max_completion_tokens` and `reasoning_effort`. |
131
139
  | `groq` | Groq | openai | `https://api.groq.com/openai/v1` | bearer | | Sends `max_completion_tokens` and `reasoning_effort`. |
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "foldrun",
3
- "version": "13.2.0",
3
+ "version": "14.0.1",
4
4
  "description": "Agents are just folders. Write them, check them, run them — on your machine.",
5
5
  "license": "FSL-1.1-ALv2",
6
6
  "type": "module",
@@ -11,7 +11,7 @@
11
11
  "node": ">=22"
12
12
  },
13
13
  "dependencies": {
14
- "@foldrun/core": "^13.1.1"
14
+ "@foldrun/core": "^14.0.1"
15
15
  },
16
16
  "files": [
17
17
  "bin",