@draht/pods 2026.3.2-7 → 2026.3.2-9
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/README.md +18 -18
- package/package.json +3 -3
package/README.md
CHANGED
|
@@ -5,7 +5,7 @@ Deploy and manage LLMs on GPU pods with automatic vLLM configuration for agentic
|
|
|
5
5
|
## Installation
|
|
6
6
|
|
|
7
7
|
```bash
|
|
8
|
-
|
|
8
|
+
bun add -g @draht/coding-agent
|
|
9
9
|
```
|
|
10
10
|
|
|
11
11
|
## What is pi?
|
|
@@ -22,7 +22,7 @@ npm install -g @mariozechner/pi
|
|
|
22
22
|
```bash
|
|
23
23
|
# Set required environment variables
|
|
24
24
|
export HF_TOKEN=your_huggingface_token # Get from https://huggingface.co/settings/tokens
|
|
25
|
-
export
|
|
25
|
+
export DRAHT_API_KEY=your_api_key # Any string you want for API authentication
|
|
26
26
|
|
|
27
27
|
# Setup a DataCrunch pod with NFS storage (models path auto-extracted)
|
|
28
28
|
pi pods setup dc1 "ssh root@1.2.3.4" \
|
|
@@ -39,7 +39,7 @@ pi agent qwen -i
|
|
|
39
39
|
|
|
40
40
|
# Use with any OpenAI-compatible client
|
|
41
41
|
export OPENAI_BASE_URL='http://1.2.3.4:8001/v1'
|
|
42
|
-
export OPENAI_API_KEY=$
|
|
42
|
+
export OPENAI_API_KEY=$DRAHT_API_KEY
|
|
43
43
|
```
|
|
44
44
|
|
|
45
45
|
## Prerequisites
|
|
@@ -121,10 +121,10 @@ pi agent <name> -i # Interactive chat mode
|
|
|
121
121
|
pi agent <name> -i -c # Continue previous session
|
|
122
122
|
|
|
123
123
|
# Standalone OpenAI-compatible agent (works with any API)
|
|
124
|
-
|
|
125
|
-
|
|
126
|
-
|
|
127
|
-
|
|
124
|
+
draht --base-url http://localhost:8000/v1 --model llama-3.1 "Hello"
|
|
125
|
+
draht --api-key sk-... "What is 2+2?" # Uses OpenAI by default
|
|
126
|
+
draht --json "What is 2+2?" # Output event stream as JSONL
|
|
127
|
+
draht -i # Interactive mode
|
|
128
128
|
```
|
|
129
129
|
|
|
130
130
|
The agent includes tools for file operations (read, list, bash, glob, rg) to test agentic capabilities, particularly useful for code navigation and analysis tasks.
|
|
@@ -311,29 +311,29 @@ response = client.chat.completions.create(
|
|
|
311
311
|
`pi` includes a standalone OpenAI-compatible agent that can work with any API:
|
|
312
312
|
|
|
313
313
|
```bash
|
|
314
|
-
# Install globally to get
|
|
315
|
-
|
|
314
|
+
# Install globally to get draht command
|
|
315
|
+
bun add -g @draht/coding-agent
|
|
316
316
|
|
|
317
317
|
# Use with OpenAI
|
|
318
|
-
|
|
318
|
+
draht --api-key sk-... "What is machine learning?"
|
|
319
319
|
|
|
320
320
|
# Use with local vLLM
|
|
321
|
-
|
|
321
|
+
draht --base-url http://localhost:8000/v1 \
|
|
322
322
|
--model meta-llama/Llama-3.1-8B-Instruct \
|
|
323
323
|
--api-key dummy \
|
|
324
324
|
"Explain quantum computing"
|
|
325
325
|
|
|
326
326
|
# Interactive mode
|
|
327
|
-
|
|
327
|
+
draht -i
|
|
328
328
|
|
|
329
329
|
# Continue previous session
|
|
330
|
-
|
|
330
|
+
draht --continue "Follow up question"
|
|
331
331
|
|
|
332
332
|
# Custom system prompt
|
|
333
|
-
|
|
333
|
+
draht --system-prompt "You are a Python expert" "Write a web scraper"
|
|
334
334
|
|
|
335
335
|
# Use responses API (for GPT-OSS models)
|
|
336
|
-
|
|
336
|
+
draht --api responses --model openai/gpt-oss-20b "Hello"
|
|
337
337
|
```
|
|
338
338
|
|
|
339
339
|
The agent supports:
|
|
@@ -411,7 +411,7 @@ Events are automatically converted to the appropriate API format (Chat Completio
|
|
|
411
411
|
|
|
412
412
|
Use `--json` flag to output the event stream as JSONL (JSON Lines) for programmatic consumption:
|
|
413
413
|
```bash
|
|
414
|
-
|
|
414
|
+
draht --api-key sk-... --json "What is 2+2?"
|
|
415
415
|
```
|
|
416
416
|
|
|
417
417
|
Each line is a complete JSON object representing an event:
|
|
@@ -502,8 +502,8 @@ ls -la ~/.pi/sessions/
|
|
|
502
502
|
## Environment Variables
|
|
503
503
|
|
|
504
504
|
- `HF_TOKEN` - HuggingFace token for model downloads
|
|
505
|
-
- `
|
|
506
|
-
- `
|
|
505
|
+
- `DRAHT_API_KEY` - API key for vLLM endpoints
|
|
506
|
+
- `DRAHT_CONFIG_DIR` - Config directory (default: `~/.pi`)
|
|
507
507
|
- `OPENAI_API_KEY` - Used by `pi-agent` when no `--api-key` provided
|
|
508
508
|
|
|
509
509
|
## License
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@draht/pods",
|
|
3
|
-
"version": "2026.3.2-
|
|
3
|
+
"version": "2026.3.2-9",
|
|
4
4
|
"description": "CLI tool for managing vLLM deployments on GPU pods",
|
|
5
5
|
"type": "module",
|
|
6
6
|
"bin": {
|
|
@@ -26,14 +26,14 @@
|
|
|
26
26
|
"license": "MIT",
|
|
27
27
|
"repository": {
|
|
28
28
|
"type": "git",
|
|
29
|
-
"url": "git+https://github.com/
|
|
29
|
+
"url": "git+https://github.com/draht-dev/draht.git",
|
|
30
30
|
"directory": "packages/pods"
|
|
31
31
|
},
|
|
32
32
|
"engines": {
|
|
33
33
|
"node": ">=20.0.0"
|
|
34
34
|
},
|
|
35
35
|
"dependencies": {
|
|
36
|
-
"@draht/agent-core": "2026.3.2-
|
|
36
|
+
"@draht/agent-core": "2026.3.2-9",
|
|
37
37
|
"chalk": "^5.5.0"
|
|
38
38
|
},
|
|
39
39
|
"devDependencies": {}
|