1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
|
# Hexai
Hexai, the AI LSP for the Helix editor.
At the moment this project is only in the proof of concept phase.
## LLM provider
Hexai exposes a simple LLM provider interface. It supports OpenAI and a local
Ollama server. Provider selection and models are configured via environment
variables.
### Selecting a provider
- Set `HEXAI_LLM_PROVIDER` to `openai` or `ollama` to force a provider.
- If not set, Hexai auto‑detects:
- Uses OpenAI when `OPENAI_API_KEY` is present.
- Uses Ollama when any `OLLAMA_*` variables are present.
- Otherwise, Hexai falls back to a basic, local completion.
### OpenAI configuration
- Required: `OPENAI_API_KEY` — your OpenAI API key.
- Optional: `OPENAI_MODEL` — model name (default: `gpt-4o-mini`).
- Optional: `OPENAI_BASE_URL` — override the API base (e.g., a compatible endpoint).
### Ollama configuration (local)
- Optional: `OLLAMA_MODEL` — model name/tag (default: `qwen2.5-coder:latest`).
- Optional: `OLLAMA_BASE_URL` or `OLLAMA_HOST` — base URL to Ollama
(default: `http://localhost:11434`).
Notes:
- For Ollama, ensure the model is available locally (e.g., `ollama pull qwen2.5-coder:latest`).
- If you run Ollama in OpenAI‑compatible mode, you may alternatively use the
OpenAI provider with `OPENAI_BASE_URL` pointing to your local endpoint.
## CLI usage and configuration
- Run LSP server over stdio:
- `hexai`
- Completion settings:
- `-max-tokens`: maximum tokens for LLM completions. If the flag isn’t provided, `HEXAI_MAX_TOKENS` is used when set.
- `-context-mode`: how much additional context to include with completion prompts (If the flag isn’t provided, `HEXAI_CONTEXT_MODE` is used when set). One of:
- `minimal`: no extra context
- `window`: include a sliding window around the cursor
- `file-on-new-func`ude the full file only when defining a new function (cursor before the opening `{`)
- `always-full`: always include the full file (may be slower/costly)
- `-context-window-lines`: line count for the sliding window when `context-mode=window`.
- `-max-context-tokens`: budget for additional context tokens. If the flag isn’t provided, `HEXAI_MAX_CONTEXT_TOKENS` is used when set.
- `-provider`: LLM provider override: `openai` or `ollama` (overrides `HEXAI_LLM_PROVIDER`).
Notes:
- Token estimation for truncation uses a simple 4 chars/token heuristic.
- Full-file context is only included by default when defining a new function to balance quality, latency, and cost.
### Flags quick reference
| Flag | Env override | Description |
|-------------------------|----------------------------|----------------------------------------------------|
| `-stdio` | — | Run as LSP over stdio (only supported mode). |
| `-log` | — | Path to log file (optional). |
| `-max-tokens` | `HEXAI_MAX_TOKENS` | Max tokens for LLM completions. |
| `-context-mode` | `HEXAI_CONTEXT_MODE` | `minimal` `window` `file-on-new-func` `always-full` |
| `-context-window-lines` | `HEXAI_CONTEXT_WINDOW_LINES` | Lines around cursor when using `window` mode. |
| `-max-context-tokens` | `HEXAI_MAX_CONTEXT_TOKENS` | Token budget for additional context. |
| `-log-preview-limit` | `HEXAI_LOG_PREVIEW_LIMIT` | Limit characters shown in LLM preview logs. |
| `-no-disk-io` | `HEXAI_NO_DISK_IO` | Disallow any disk reads for context. |
| `-provider` | `HEXAI_LLM_PROVIDER` | Force LLM provider: `openai` or `ollama`. |
### Environment quick reference (providers)
- `HEXAI_LLM_PROVIDER`: `openai` | `ollama` (optional; otherwise auto‑detect).
- OpenAI: `OPENAI_API_KEY` (required), `OPENAI_MODEL`, `OPENAI_BASE_URL`.
- Ollama: `OLLAMA_MODEL`, `OLLAMA_BASE_URL` or `OLLAMA_HOST`.
|
