@inference-gateway/cli 0.140.1 → 0.141.0
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.
- package/README.md +1 -231
- package/package.json +1 -1
package/README.md
CHANGED
|
@@ -25,7 +25,6 @@ An agentic command-line assistant that writes code, understands project context,
|
|
|
25
25
|
- [Features](#features)
|
|
26
26
|
- [Installation](#installation)
|
|
27
27
|
- [Quick Start](#quick-start)
|
|
28
|
-
- [Claude Code Mode (Subscription)](#claude-code-mode-subscription)
|
|
29
28
|
- [Commands](#commands)
|
|
30
29
|
- [Tools for LLMs](#tools-for-llms)
|
|
31
30
|
- [Configuration](#configuration)
|
|
@@ -312,236 +311,6 @@ Now that you're up and running, explore these guides:
|
|
|
312
311
|
- **[Shortcuts Guide](docs/shortcuts-guide.md)** - Custom shortcuts and AI-powered snippets
|
|
313
312
|
- **[A2A Agents](docs/agents-configuration.md)** - Agent-to-agent communication setup
|
|
314
313
|
|
|
315
|
-
## Claude Code Mode (Subscription)
|
|
316
|
-
|
|
317
|
-
Save on API costs by using your Claude Max or Pro subscription instead of pay-as-you-go API pricing.
|
|
318
|
-
|
|
319
|
-
### Overview
|
|
320
|
-
|
|
321
|
-
Claude Code mode enables you to use your **Claude Max or Pro subscription** ($100-200/month fixed cost)
|
|
322
|
-
instead of paying per token via the Anthropic API. This is ideal for heavy users who want
|
|
323
|
-
predictable monthly costs.
|
|
324
|
-
|
|
325
|
-
**Cost Comparison:**
|
|
326
|
-
|
|
327
|
-
| Mode | Pricing | Best For |
|
|
328
|
-
| -------------------- | --------------------------------------- | --------------------------------------- |
|
|
329
|
-
| **Gateway Mode** | Pay per token ($3-$75 per million) | API users, multi-provider needs |
|
|
330
|
-
| **Claude Code Mode** | Fixed monthly ($100-200) | Heavy Claude users, cost predictability |
|
|
331
|
-
|
|
332
|
-
### Prerequisites
|
|
333
|
-
|
|
334
|
-
- **Claude Max or Pro subscription** - Required ($100-200/month)
|
|
335
|
-
- **Claude Code CLI** - Official CLI from Anthropic
|
|
336
|
-
|
|
337
|
-
Install the Claude Code CLI:
|
|
338
|
-
|
|
339
|
-
```bash
|
|
340
|
-
npm install -g @anthropic-ai/claude-code
|
|
341
|
-
```
|
|
342
|
-
|
|
343
|
-
### Setup
|
|
344
|
-
|
|
345
|
-
1. **Configure for Claude Code mode**:
|
|
346
|
-
|
|
347
|
-
Edit `.infer/config.yaml`:
|
|
348
|
-
|
|
349
|
-
```yaml
|
|
350
|
-
# Enable Claude Code mode
|
|
351
|
-
claude_code:
|
|
352
|
-
enabled: true
|
|
353
|
-
cli_path: claude # or /usr/local/bin/claude if not in PATH
|
|
354
|
-
timeout: 600
|
|
355
|
-
max_output_tokens: 32000
|
|
356
|
-
thinking_budget: 10000
|
|
357
|
-
|
|
358
|
-
# Disable gateway mode
|
|
359
|
-
gateway:
|
|
360
|
-
run: false
|
|
361
|
-
|
|
362
|
-
# Set model (no provider prefix needed)
|
|
363
|
-
agent:
|
|
364
|
-
model: claude-sonnet-4-5-20250929
|
|
365
|
-
```
|
|
366
|
-
|
|
367
|
-
2. **Authenticate with your subscription**:
|
|
368
|
-
|
|
369
|
-
```bash
|
|
370
|
-
infer claude-code setup
|
|
371
|
-
```
|
|
372
|
-
|
|
373
|
-
This opens your browser to authenticate with your Claude Max/Pro account.
|
|
374
|
-
|
|
375
|
-
3. **Verify authentication**:
|
|
376
|
-
|
|
377
|
-
```bash
|
|
378
|
-
infer claude-code test
|
|
379
|
-
```
|
|
380
|
-
|
|
381
|
-
4. **Use normally**:
|
|
382
|
-
|
|
383
|
-
```bash
|
|
384
|
-
infer chat # Now using your subscription!
|
|
385
|
-
```
|
|
386
|
-
|
|
387
|
-
### Available Commands
|
|
388
|
-
|
|
389
|
-
- `infer claude-code setup` - Authenticate with Claude subscription
|
|
390
|
-
- `infer claude-code test` - Test authentication and CLI integration
|
|
391
|
-
|
|
392
|
-
### Configuration Options
|
|
393
|
-
|
|
394
|
-
```yaml
|
|
395
|
-
claude_code:
|
|
396
|
-
enabled: true # Enable/disable Claude Code mode
|
|
397
|
-
cli_path: claude # Path to claude binary
|
|
398
|
-
timeout: 600 # Command timeout in seconds
|
|
399
|
-
max_output_tokens: 32000 # Maximum output tokens per request
|
|
400
|
-
thinking_budget: 10000 # Token budget for extended thinking
|
|
401
|
-
extra_args: # Extra arguments appended verbatim to the claude CLI invocation
|
|
402
|
-
- --max-turns
|
|
403
|
-
- "5"
|
|
404
|
-
```
|
|
405
|
-
|
|
406
|
-
**Environment Variables:**
|
|
407
|
-
|
|
408
|
-
```bash
|
|
409
|
-
export INFER_CLAUDE_CODE_ENABLED=true
|
|
410
|
-
export INFER_CLAUDE_CODE_CLI_PATH=/usr/local/bin/claude
|
|
411
|
-
export INFER_CLAUDE_CODE_TIMEOUT=600
|
|
412
|
-
export INFER_CLAUDE_CODE_EXTRA_ARGS="--max-turns,5" # comma/newline-separated; wins over --claude-code-extra-args
|
|
413
|
-
```
|
|
414
|
-
|
|
415
|
-
**Pass-through behavior:**
|
|
416
|
-
|
|
417
|
-
Claude Code mode is a pure pass-through: infer does not inject its system prompt, context blocks, or
|
|
418
|
-
system reminders, and does not re-execute claude's tool calls locally - claude runs with its own
|
|
419
|
-
defaults and native tools. Infer's `prompts.yaml` and `reminders.yaml` do not apply in this mode.
|
|
420
|
-
|
|
421
|
-
To add instructions on top of claude's built-in system prompt (passed via `--append-system-prompt`),
|
|
422
|
-
set the dedicated prompt in `.infer/prompts.yaml` (empty by default):
|
|
423
|
-
|
|
424
|
-
```yaml
|
|
425
|
-
agent:
|
|
426
|
-
system_prompt_claude_code: "Always answer in English."
|
|
427
|
-
```
|
|
428
|
-
|
|
429
|
-
Or via environment variable:
|
|
430
|
-
|
|
431
|
-
```bash
|
|
432
|
-
export INFER_PROMPTS_AGENT_SYSTEM_PROMPT_CLAUDE_CODE="Always answer in English."
|
|
433
|
-
```
|
|
434
|
-
|
|
435
|
-
### Features and Limitations
|
|
436
|
-
|
|
437
|
-
| Feature | Gateway Mode | Claude Code Mode |
|
|
438
|
-
| --------------------- | --------------------------------------- | ---------------------------------------- |
|
|
439
|
-
| **Cost** | Pay-per-token | Fixed monthly |
|
|
440
|
-
| **Providers** | All providers (Anthropic, OpenAI, etc.) | Claude only |
|
|
441
|
-
| **Models** | All provider models | Claude models only |
|
|
442
|
-
| **Images** | ✓ Supported | ✗ Not supported (stripped from messages) |
|
|
443
|
-
| **Prompt Caching** | ✓ Supported | ✗ Not available via CLI |
|
|
444
|
-
| **Streaming** | ✓ Supported | ✓ Supported |
|
|
445
|
-
| **Tool Execution** | ✓ Supported | ✓ Supported |
|
|
446
|
-
| **Extended Thinking** | ✓ Supported | ✓ Supported |
|
|
447
|
-
| **Authentication** | API keys | Browser login |
|
|
448
|
-
|
|
449
|
-
**Supported Models:**
|
|
450
|
-
|
|
451
|
-
The following Claude models are available via Claude Code subscription mode:
|
|
452
|
-
|
|
453
|
-
**Claude 4.5 Series (Latest):**
|
|
454
|
-
|
|
455
|
-
- `claude-opus-4-5` - Most capable Claude model (vision support)
|
|
456
|
-
- `claude-haiku-4-5-20251001` - Fastest Claude model (vision support)
|
|
457
|
-
- `claude-sonnet-4-5-20250929` - Latest Sonnet model (default, vision support)
|
|
458
|
-
|
|
459
|
-
**Claude 4.1 Series:**
|
|
460
|
-
|
|
461
|
-
- `claude-opus-4-1-20250805` - Claude 4.1 Opus (vision support)
|
|
462
|
-
- `claude-sonnet-4-1-20250805` - Claude 4.1 Sonnet (vision support)
|
|
463
|
-
|
|
464
|
-
**Claude 4 Series:**
|
|
465
|
-
|
|
466
|
-
- `claude-opus-4-20250514` - Claude 4 Opus (vision support)
|
|
467
|
-
- `claude-sonnet-4-20250514` - Claude 4 Sonnet (vision support)
|
|
468
|
-
|
|
469
|
-
**Claude 3.7 Series:**
|
|
470
|
-
|
|
471
|
-
- `claude-3-7-sonnet-20250219` - Claude 3.7 Sonnet (vision support)
|
|
472
|
-
|
|
473
|
-
**Claude 3.5 Series:**
|
|
474
|
-
|
|
475
|
-
- `claude-3-5-haiku-20241022` - Claude 3.5 Haiku (vision support)
|
|
476
|
-
|
|
477
|
-
**Claude 3 Series:**
|
|
478
|
-
|
|
479
|
-
- `claude-3-haiku-20240307` - Claude 3 Haiku (vision support)
|
|
480
|
-
- `claude-3-opus-20240229` - Claude 3 Opus (vision support)
|
|
481
|
-
|
|
482
|
-
**Note:** All modern Claude models support vision capabilities. The Claude Code CLI automatically strips images
|
|
483
|
-
from messages when using subscription mode.
|
|
484
|
-
|
|
485
|
-
### Troubleshooting
|
|
486
|
-
|
|
487
|
-
**CLI Not Found:**
|
|
488
|
-
|
|
489
|
-
```bash
|
|
490
|
-
# Check if Claude CLI is installed
|
|
491
|
-
which claude
|
|
492
|
-
|
|
493
|
-
# If not found, install it
|
|
494
|
-
npm install -g @anthropic-ai/claude-code
|
|
495
|
-
|
|
496
|
-
# Or set custom path in config
|
|
497
|
-
claude_code:
|
|
498
|
-
cli_path: /full/path/to/claude
|
|
499
|
-
```
|
|
500
|
-
|
|
501
|
-
**Authentication Issues:**
|
|
502
|
-
|
|
503
|
-
```bash
|
|
504
|
-
# Re-authenticate
|
|
505
|
-
infer claude-code setup
|
|
506
|
-
|
|
507
|
-
# Test authentication
|
|
508
|
-
infer claude-code test
|
|
509
|
-
```
|
|
510
|
-
|
|
511
|
-
**Update CLI:**
|
|
512
|
-
|
|
513
|
-
```bash
|
|
514
|
-
npm update -g @anthropic-ai/claude-code
|
|
515
|
-
```
|
|
516
|
-
|
|
517
|
-
### Switching Between Modes
|
|
518
|
-
|
|
519
|
-
You can easily switch between gateway and Claude Code modes:
|
|
520
|
-
|
|
521
|
-
**To Claude Code mode:**
|
|
522
|
-
|
|
523
|
-
```yaml
|
|
524
|
-
# .infer/config.yaml
|
|
525
|
-
claude_code:
|
|
526
|
-
enabled: true
|
|
527
|
-
gateway:
|
|
528
|
-
run: false
|
|
529
|
-
agent:
|
|
530
|
-
model: claude-sonnet-4-5-20250929 # No provider prefix
|
|
531
|
-
```
|
|
532
|
-
|
|
533
|
-
**To Gateway mode:**
|
|
534
|
-
|
|
535
|
-
```yaml
|
|
536
|
-
# .infer/config.yaml
|
|
537
|
-
claude_code:
|
|
538
|
-
enabled: false
|
|
539
|
-
gateway:
|
|
540
|
-
run: true
|
|
541
|
-
agent:
|
|
542
|
-
model: anthropic/claude-sonnet-4-5-20250929 # With provider prefix
|
|
543
|
-
```
|
|
544
|
-
|
|
545
314
|
## Commands
|
|
546
315
|
|
|
547
316
|
The CLI provides several commands for different workflows. For detailed documentation, see [Commands Reference](docs/commands-reference.md).
|
|
@@ -906,6 +675,7 @@ All configuration can be set via environment variables with the `INFER_` prefix:
|
|
|
906
675
|
|
|
907
676
|
```bash
|
|
908
677
|
export INFER_GATEWAY_URL="http://localhost:8080"
|
|
678
|
+
export INFER_GATEWAY_MOCK=true # embedded mock gateway with canned scenario responses, no real LLM
|
|
909
679
|
export INFER_AGENT_MODEL="deepseek/deepseek-v4-pro"
|
|
910
680
|
export INFER_TOOLS_BASH_ENABLED=true
|
|
911
681
|
export INFER_CHAT_THEME="tokyo-night"
|