@bilalimamoglu/sift 0.3.0 → 0.3.1

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/README.md CHANGED
@@ -49,6 +49,13 @@ That writes a machine-wide config to:
49
49
 
50
50
  After that, any terminal on the machine can use `sift`. A repo-local config can still override it later.
51
51
 
52
+ To switch between saved native providers without editing YAML:
53
+
54
+ ```bash
55
+ sift config use openai
56
+ sift config use openrouter
57
+ ```
58
+
52
59
  If you prefer manual setup, this is the smallest useful OpenAI setup:
53
60
 
54
61
  ```bash
@@ -58,6 +65,13 @@ export SIFT_MODEL=gpt-5-nano
58
65
  export OPENAI_API_KEY=your_openai_api_key
59
66
  ```
60
67
 
68
+ If you prefer manual setup, this is the smallest useful OpenRouter setup:
69
+
70
+ ```bash
71
+ export SIFT_PROVIDER=openrouter
72
+ export OPENROUTER_API_KEY=your_openrouter_api_key
73
+ ```
74
+
61
75
  Then check it:
62
76
 
63
77
  ```bash
@@ -130,6 +144,8 @@ sift exec --preset test-status -- <test command>
130
144
 
131
145
  If `standard` already names the main failure buckets, counts, and hints, stop there and read code.
132
146
 
147
+ If `standard` still includes an unknown bucket or ends with `Decision: zoom`, do one deeper sift pass before you fall back to raw traceback.
148
+
133
149
  Then use this order:
134
150
  1. `sift exec --preset test-status -- <test command>`
135
151
  2. `sift rerun`
@@ -148,6 +164,8 @@ Most of the time, you do not need JSON. Start with text first.
148
164
 
149
165
  If `standard` already shows bucket-level root cause, `Anchor`, and `Fix`, do not re-verify the same bucket with raw pytest. At most do one targeted source read before you edit.
150
166
 
167
+ If diagnose output still contains an unknown bucket or `Decision: zoom`, take one sift zoom step before raw traceback.
168
+
151
169
  Use diagnose JSON only when automation or machine branching really needs it:
152
170
 
153
171
  ```bash
@@ -378,6 +396,7 @@ Useful commands:
378
396
 
379
397
  ```bash
380
398
  sift config setup
399
+ sift config use openrouter
381
400
  sift config init
382
401
  sift config show
383
402
  sift config validate
@@ -430,10 +449,14 @@ runtime:
430
449
  rawFallback: true
431
450
  ```
432
451
 
433
- ## OpenAI vs OpenAI-compatible
452
+ ## OpenAI vs OpenRouter vs OpenAI-compatible
434
453
 
435
454
  Use `provider: openai` for `api.openai.com`.
436
455
 
456
+ Use `provider: openrouter` for the native OpenRouter path. It defaults to:
457
+ - `baseUrl: https://openrouter.ai/api/v1`
458
+ - `model: openrouter/free`
459
+
437
460
  Use `provider: openai-compatible` for third-party compatible gateways or self-hosted endpoints.
438
461
 
439
462
  For OpenAI:
@@ -441,6 +464,11 @@ For OpenAI:
441
464
  export OPENAI_API_KEY=your_openai_api_key
442
465
  ```
443
466
 
467
+ For OpenRouter:
468
+ ```bash
469
+ export OPENROUTER_API_KEY=your_openrouter_api_key
470
+ ```
471
+
444
472
  For third-party compatible endpoints, use either the endpoint-native env var or:
445
473
 
446
474
  ```bash