-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy path.env.example
702 lines (657 loc) · 36.6 KB
/
.env.example
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
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
# Example .env config, version: development
# The following example contains all currently supported environment parameters.
# They are used to set up the connection to LLM and configure it for various operations. Your actual configuration may be significantly smaller.
# Some options are commented out, you can uncomment them to customize the behavior in special cases - it will take priority only for those specific cases.
# Configuration file should be named ".env" and it can be placed to the following locations:
# Project local config: <Project root>/.perpetual/.env
# Global config. On Linux: ~/.config/Perpetual/.env ; On Windows: <User profile dir>\AppData\Roaming\Perpetual\.env
# Also, the parameters can be exported to the system environment before running the utility, then they will have priority over the parameters in the configuration files.
# Provider selection for particular operations and stages
# You can offload tasks to different providers to balance between generation quality and costs.
# LLM_PROVIDER_OP_ANNOTATE="anthropic"
# LLM_PROVIDER_OP_ANNOTATE_POST="anthropic"
# LLM_PROVIDER_OP_IMPLEMENT_STAGE1="anthropic"
# LLM_PROVIDER_OP_IMPLEMENT_STAGE2="anthropic"
# LLM_PROVIDER_OP_IMPLEMENT_STAGE3="anthropic"
# LLM_PROVIDER_OP_IMPLEMENT_STAGE4="anthropic"
# LLM_PROVIDER_OP_DOC_STAGE1="anthropic"
# LLM_PROVIDER_OP_DOC_STAGE2="anthropic"
# LLM_PROVIDER_OP_EXPLAIN_STAGE1="anthropic"
# LLM_PROVIDER_OP_EXPLAIN_STAGE2="anthropic"
# Default, fallback provider selection, will be used if options above are not used
LLM_PROVIDER="anthropic"
# LLM_PROVIDER="openai"
# LLM_PROVIDER="ollama"
# LLM_PROVIDER="generic"
# NOTE: you can also setup multiple profiles for supported LLM providers using following naming scheme: <PROVIDER><PROFILE NUMBER>_<OPTION>
# examples:
# LLM_PROVIDER="ollama1"
# OLLAMA1_BASE_URL=...
# OLLAMA1_MODEL=...
# LLM_PROVIDER="generic1"
# GENERIC1_BASE_URL=...
# GENERIC1_MODEL=...
# Options for Anthropic provider. Below are sane defaults for Anthropic provider (as of Jan 2025)
ANTHROPIC_API_KEY="<your api key goes here>"
ANTHROPIC_BASE_URL="https://api.anthropic.com/v1"
ANTHROPIC_MODEL_OP_ANNOTATE="claude-3-haiku-20240307"
ANTHROPIC_MODEL_OP_ANNOTATE_POST="claude-3-haiku-20240307" # used to process multiple response-variants if any
# ANTHROPIC_MODEL_OP_IMPLEMENT_STAGE1="claude-3-7-sonnet-latest"
# ANTHROPIC_MODEL_OP_IMPLEMENT_STAGE2="claude-3-7-sonnet-latest"
# ANTHROPIC_MODEL_OP_IMPLEMENT_STAGE3="claude-3-7-sonnet-latest"
# ANTHROPIC_MODEL_OP_IMPLEMENT_STAGE4="claude-3-7-sonnet-latest"
# ANTHROPIC_MODEL_OP_DOC_STAGE1="claude-3-7-sonnet-latest"
ANTHROPIC_MODEL_OP_DOC_STAGE2="claude-3-7-sonnet-latest"
# ANTHROPIC_MODEL_OP_EXPLAIN_STAGE1="claude-3-7-sonnet-latest"
ANTHROPIC_MODEL_OP_EXPLAIN_STAGE2="claude-3-7-sonnet-latest"
ANTHROPIC_MODEL="claude-3-7-sonnet-latest"
ANTHROPIC_VARIANT_COUNT_OP_ANNOTATE="1" # how much annotate-response variants to generate
ANTHROPIC_VARIANT_SELECTION_OP_ANNOTATE="short" # how to select final variant: short, long, combine, best
ANTHROPIC_VARIANT_COUNT="1" # will be used as fallback
ANTHROPIC_VARIANT_SELECTION="short" # will be used as fallback
# Switch to use structured JSON output format for supported operations, supported values: plain, json. Default: plain
# The "plain" method seems to work better here, since it uses XML-style tags, for which anthropic models were initially better trained than for JSON.
# ANTHROPIC_FORMAT_OP_IMPLEMENT_STAGE1="json"
# ANTHROPIC_FORMAT_OP_IMPLEMENT_STAGE3="json"
# ANTHROPIC_FORMAT_OP_DOC_STAGE1="json"
# ANTHROPIC_FORMAT_OP_EXPLAIN_STAGE1="json"
ANTHROPIC_MAX_TOKENS_OP_ANNOTATE="768"
ANTHROPIC_MAX_TOKENS_OP_ANNOTATE_POST="768"
ANTHROPIC_MAX_TOKENS_OP_IMPLEMENT_STAGE1="512" # file-list for review, long list is probably an error
ANTHROPIC_MAX_TOKENS_OP_IMPLEMENT_STAGE2="1536" # work plan also should not be too big
ANTHROPIC_MAX_TOKENS_OP_IMPLEMENT_STAGE3="512" # file-list for processing, long list is probably an error
ANTHROPIC_MAX_TOKENS_OP_IMPLEMENT_STAGE4="8192" # generated code output limit should be as big as possible
ANTHROPIC_MAX_TOKENS_OP_DOC_STAGE1="768" # file-list for review, long list is probably an error
ANTHROPIC_MAX_TOKENS_OP_DOC_STAGE2="32768" # generated document output limit should be big
ANTHROPIC_MAX_TOKENS_OP_EXPLAIN_STAGE1="512" # file-list for review
ANTHROPIC_MAX_TOKENS_OP_EXPLAIN_STAGE2="32768" # generated answer output limit
ANTHROPIC_MAX_TOKENS="4096" # default limit
ANTHROPIC_MAX_TOKENS_SEGMENTS="3"
# Extended thinking. Should work with newer models. 1024 is a minimum for claude 3.7 sonnet.
# May be incompatible with some other parameters (temperature)
# If set to 0 - explicitly disable extra thinking in api call,
# if > 0 enable thinking block in api call and set budget_tokens,
# If unset - do not alter api call and response in any way
# ANTHROPIC_THINK_TOKENS_OP_ANNOTATE="0"
# ANTHROPIC_THINK_TOKENS_OP_ANNOTATE_POST="0"
# ANTHROPIC_THINK_TOKENS_OP_IMPLEMENT_STAGE1="0" # file list
# ANTHROPIC_THINK_TOKENS_OP_IMPLEMENT_STAGE2="2048" # work plan
# ANTHROPIC_THINK_TOKENS_OP_IMPLEMENT_STAGE3="0" # file list
# ANTHROPIC_THINK_TOKENS_OP_IMPLEMENT_STAGE4="1024" # code implementation
# ANTHROPIC_THINK_TOKENS_OP_DOC_STAGE1="0" # file list
ANTHROPIC_THINK_TOKENS_OP_DOC_STAGE2="4096" # document process
# ANTHROPIC_THINK_TOKENS_OP_EXPLAIN_STAGE1="0" # file list
ANTHROPIC_THINK_TOKENS_OP_EXPLAIN_STAGE2="4096" # answer generation
ANTHROPIC_THINK_TOKENS="0"
ANTHROPIC_ON_FAIL_RETRIES_OP_ANNOTATE="1"
# ANTHROPIC_ON_FAIL_RETRIES_OP_IMPLEMENT_STAGE1="3"
# ANTHROPIC_ON_FAIL_RETRIES_OP_IMPLEMENT_STAGE2="3"
# ANTHROPIC_ON_FAIL_RETRIES_OP_IMPLEMENT_STAGE3="3"
# ANTHROPIC_ON_FAIL_RETRIES_OP_IMPLEMENT_STAGE4="3"
# ANTHROPIC_ON_FAIL_RETRIES_OP_DOC_STAGE1="3"
# ANTHROPIC_ON_FAIL_RETRIES_OP_DOC_STAGE2="3"
# ANTHROPIC_ON_FAIL_RETRIES_OP_EXPLAIN_STAGE1="3"
# ANTHROPIC_ON_FAIL_RETRIES_OP_EXPLAIN_STAGE2="3"
ANTHROPIC_ON_FAIL_RETRIES="3"
# ANTHROPIC_TEMPERATURE_OP_ANNOTATE="0.5"
# ANTHROPIC_TEMPERATURE_OP_ANNOTATE_POST="0.5"
ANTHROPIC_TEMPERATURE_OP_IMPLEMENT_STAGE1="0.2" # less creative for file-list output
# ANTHROPIC_TEMPERATURE_OP_IMPLEMENT_STAGE2="0.5"
ANTHROPIC_TEMPERATURE_OP_IMPLEMENT_STAGE3="0.2" # less creative for file-list output
# ANTHROPIC_TEMPERATURE_OP_IMPLEMENT_STAGE4="0.5"
ANTHROPIC_TEMPERATURE_OP_DOC_STAGE1="0.2" # less creative for file-list output
ANTHROPIC_TEMPERATURE_OP_DOC_STAGE2="1" # value 1 for needed thinking model
ANTHROPIC_TEMPERATURE_OP_EXPLAIN_STAGE1="0.2" # less creative for file-list output
ANTHROPIC_TEMPERATURE_OP_EXPLAIN_STAGE2="1" # value 1 for needed thinking model
ANTHROPIC_TEMPERATURE="0.5"
# Advanced options that currently supported with Anthropic. You mostly not need to use them
# ANTHROPIC_TOP_K_OP_ANNOTATE="40"
# ANTHROPIC_TOP_K_OP_ANNOTATE_POST="40"
# ANTHROPIC_TOP_K_OP_IMPLEMENT_STAGE1="40"
# ANTHROPIC_TOP_K_OP_IMPLEMENT_STAGE2="40"
# ANTHROPIC_TOP_K_OP_IMPLEMENT_STAGE3="40"
# ANTHROPIC_TOP_K_OP_IMPLEMENT_STAGE4="40"
# ANTHROPIC_TOP_K_OP_DOC_STAGE1="40"
# ANTHROPIC_TOP_K_OP_DOC_STAGE2="40"
# ANTHROPIC_TOP_K_OP_EXPLAIN_STAGE1="40"
# ANTHROPIC_TOP_K_OP_EXPLAIN_STAGE2="40"
# ANTHROPIC_TOP_K="40"
# ANTHROPIC_TOP_P_OP_ANNOTATE="0.9"
# ANTHROPIC_TOP_P_OP_ANNOTATE_POST="0.9"
# ANTHROPIC_TOP_P_OP_IMPLEMENT_STAGE1="0.9"
# ANTHROPIC_TOP_P_OP_IMPLEMENT_STAGE2="0.9"
# ANTHROPIC_TOP_P_OP_IMPLEMENT_STAGE3="0.9"
# ANTHROPIC_TOP_P_OP_IMPLEMENT_STAGE4="0.9"
# ANTHROPIC_TOP_P_OP_DOC_STAGE1="0.9"
# ANTHROPIC_TOP_P_OP_DOC_STAGE2="0.9"
# ANTHROPIC_TOP_P_OP_EXPLAIN_STAGE1="0.9"
# ANTHROPIC_TOP_P_OP_EXPLAIN_STAGE2="0.9"
# ANTHROPIC_TOP_P="0.9"
# Options for OpenAI provider. Below are sane defaults for OpenAI provider (as of Jan 2025)
OPENAI_API_KEY="<your api key goes here>"
OPENAI_BASE_URL="https://api.openai.com/v1"
OPENAI_MODEL_OP_ANNOTATE="gpt-4o-mini"
OPENAI_MODEL_OP_ANNOTATE_POST="gpt-4o-mini" # used to process multiple response-variants if any
# OPENAI_MODEL_OP_IMPLEMENT_STAGE1="gpt-4o"
# OPENAI_MODEL_OP_IMPLEMENT_STAGE2="gpt-4o"
# OPENAI_MODEL_OP_IMPLEMENT_STAGE3="gpt-4o"
# OPENAI_MODEL_OP_IMPLEMENT_STAGE4="gpt-4o"
OPENAI_MODEL_OP_DOC_STAGE1="gpt-4o"
# OPENAI_MODEL_OP_DOC_STAGE2="o1-mini-2024-09-12" # good for generating initial document structure from your draft
# OPENAI_MODEL_OP_DOC_STAGE2="gpt-4o"
OPENAI_MODEL_OP_DOC_STAGE2="gpt-4.5-preview"
# OPENAI_MODEL_OP_EXPLAIN_STAGE1="gpt-4o"
# OPENAI_MODEL_OP_EXPLAIN_STAGE2="gpt-4o"
# OPENAI_MODEL_OP_EXPLAIN_STAGE2="o1-mini-2024-09-12"
# OPENAI_MODEL_OP_EXPLAIN_STAGE2="gpt-4.5-preview"
OPENAI_MODEL="gpt-4o"
OPENAI_VARIANT_COUNT_OP_ANNOTATE="1" # how much annotate-response variants to generate
OPENAI_VARIANT_SELECTION_OP_ANNOTATE="short" # how to select final variant: short, long, combine, best
OPENAI_VARIANT_COUNT="1" # will be used as fallback
OPENAI_VARIANT_SELECTION="short" # will be used as fallback
# Switch to use structured JSON output format for supported operations
# Supported values: plain, json. Default: plain
# OPENAI_FORMAT_OP_IMPLEMENT_STAGE1="json"
# OPENAI_FORMAT_OP_IMPLEMENT_STAGE3="json"
# OPENAI_FORMAT_OP_DOC_STAGE1="json"
# OPENAI_FORMAT_OP_EXPLAIN_STAGE1="json"
OPENAI_MAX_TOKENS_OP_ANNOTATE="768"
OPENAI_MAX_TOKENS_OP_ANNOTATE_POST="768"
OPENAI_MAX_TOKENS_OP_IMPLEMENT_STAGE1="512" # file-list for review, long list is probably an error
OPENAI_MAX_TOKENS_OP_IMPLEMENT_STAGE2="1536" # work plan also should not be too big
OPENAI_MAX_TOKENS_OP_IMPLEMENT_STAGE3="512" # file-list for processing, long list is probably an error
OPENAI_MAX_TOKENS_OP_IMPLEMENT_STAGE4="16384" # generated code output limit should be as big as possible
OPENAI_MAX_TOKENS_OP_DOC_STAGE1="768" # file-list for review, long list is probably an error
OPENAI_MAX_TOKENS_OP_DOC_STAGE2="16384" # generated document output limit should be as big as possible
OPENAI_MAX_TOKENS_OP_EXPLAIN_STAGE1="512" # file-list for review
OPENAI_MAX_TOKENS_OP_EXPLAIN_STAGE2="8192" # generated answer output limit
OPENAI_MAX_TOKENS="4096" # default limit
OPENAI_MAX_TOKENS_SEGMENTS="3"
OPENAI_ON_FAIL_RETRIES_OP_ANNOTATE="1"
# OPENAI_ON_FAIL_RETRIES_OP_IMPLEMENT_STAGE1="3"
# OPENAI_ON_FAIL_RETRIES_OP_IMPLEMENT_STAGE2="3"
# OPENAI_ON_FAIL_RETRIES_OP_IMPLEMENT_STAGE3="3"
# OPENAI_ON_FAIL_RETRIES_OP_IMPLEMENT_STAGE4="3"
# OPENAI_ON_FAIL_RETRIES_OP_DOC_STAGE1="3"
# OPENAI_ON_FAIL_RETRIES_OP_DOC_STAGE2="3"
# OPENAI_ON_FAIL_RETRIES_OP_EXPLAIN_STAGE1="3"
# OPENAI_ON_FAIL_RETRIES_OP_EXPLAIN_STAGE2="3"
OPENAI_ON_FAIL_RETRIES="3"
# OPENAI_TEMPERATURE_OP_ANNOTATE="0.5"
# OPENAI_TEMPERATURE_OP_ANNOTATE_POST="0.5"
OPENAI_TEMPERATURE_OP_IMPLEMENT_STAGE1="0.2" # less creative for file-list output
# OPENAI_TEMPERATURE_OP_IMPLEMENT_STAGE2="0.5"
OPENAI_TEMPERATURE_OP_IMPLEMENT_STAGE3="0.2" # less creative for file-list output
# OPENAI_TEMPERATURE_OP_IMPLEMENT_STAGE4="0.5"
OPENAI_TEMPERATURE_OP_DOC_STAGE1="0.2" # less creative for file-list output
OPENAI_TEMPERATURE_OP_DOC_STAGE2="0.7" # more creative when writing documentation
OPENAI_TEMPERATURE_OP_EXPLAIN_STAGE1="0.2" # less creative for file-list output
OPENAI_TEMPERATURE_OP_EXPLAIN_STAGE2="0.7"
OPENAI_TEMPERATURE="0.5"
# Advanced options for finetuning. Generally you do not need them.
# OPENAI_TOP_P_OP_ANNOTATE="0.9"
# OPENAI_TOP_P_OP_ANNOTATE_POST="0.9"
# OPENAI_TOP_P_OP_IMPLEMENT_STAGE1="0.9"
# OPENAI_TOP_P_OP_IMPLEMENT_STAGE2="0.9"
# OPENAI_TOP_P_OP_IMPLEMENT_STAGE3="0.9"
# OPENAI_TOP_P_OP_IMPLEMENT_STAGE4="0.9"
# OPENAI_TOP_P_OP_DOC_STAGE1="0.9"
# OPENAI_TOP_P_OP_DOC_STAGE2="0.9"
# OPENAI_TOP_P_OP_EXPLAIN_STAGE1="0.9"
# OPENAI_TOP_P_OP_EXPLAIN_STAGE2="0.9"
# OPENAI_TOP_P="0.9"
# OPENAI_SEED_OP_ANNOTATE="42"
# OPENAI_SEED_OP_ANNOTATE_POST="42"
# OPENAI_SEED_OP_IMPLEMENT_STAGE1="42"
# OPENAI_SEED_OP_IMPLEMENT_STAGE2="42"
# OPENAI_SEED_OP_IMPLEMENT_STAGE3="42"
# OPENAI_SEED_OP_IMPLEMENT_STAGE4="42"
# OPENAI_SEED_OP_DOC_STAGE1="42"
# OPENAI_SEED_OP_DOC_STAGE2="42"
# OPENAI_SEED_OP_EXPLAIN_STAGE1="42"
# OPENAI_SEED_OP_EXPLAIN_STAGE2="42"
# OPENAI_SEED="42"
# OPENAI_REASONING_EFFORT_OP_ANNOTATE="medium" # will work only for some reasoning models like full o1 (not o1-preview or o1-mini), values: low, medium, high
# OPENAI_REASONING_EFFORT_OP_ANNOTATE_POST="medium"
# OPENAI_REASONING_EFFORT_OP_IMPLEMENT_STAGE1="medium"
# OPENAI_REASONING_EFFORT_OP_IMPLEMENT_STAGE2="medium"
# OPENAI_REASONING_EFFORT_OP_IMPLEMENT_STAGE3="medium"
# OPENAI_REASONING_EFFORT_OP_IMPLEMENT_STAGE4="medium"
# OPENAI_REASONING_EFFORT_OP_DOC_STAGE1="medium"
# OPENAI_REASONING_EFFORT_OP_DOC_STAGE2="medium"
# OPENAI_REASONING_EFFORT_OP_EXPLAIN_STAGE1="medium"
# OPENAI_REASONING_EFFORT_OP_EXPLAIN_STAGE2="medium"
# OPENAI_REASONING_EFFORT="medium"
# OPENAI_FREQ_PENALTY_OP_ANNOTATE="1.0"
# OPENAI_FREQ_PENALTY_OP_ANNOTATE_POST="1.0"
# OPENAI_FREQ_PENALTY_OP_IMPLEMENT_STAGE1="1.0"
# OPENAI_FREQ_PENALTY_OP_IMPLEMENT_STAGE2="1.0"
# OPENAI_FREQ_PENALTY_OP_IMPLEMENT_STAGE3="1.0"
# OPENAI_FREQ_PENALTY_OP_IMPLEMENT_STAGE4="1.0"
# OPENAI_FREQ_PENALTY_OP_DOC_STAGE1="1.0"
# OPENAI_FREQ_PENALTY_OP_DOC_STAGE2="1.0"
# OPENAI_FREQ_PENALTY_OP_EXPLAIN_STAGE1="1.0"
# OPENAI_FREQ_PENALTY_OP_EXPLAIN_STAGE2="1.0"
# OPENAI_FREQ_PENALTY="1.0"
# OPENAI_PRESENCE_PENALTY_OP_ANNOTATE="1.0"
# OPENAI_PRESENCE_PENALTY_OP_ANNOTATE_POST="1.0"
# OPENAI_PRESENCE_PENALTY_OP_IMPLEMENT_STAGE1="1.0"
# OPENAI_PRESENCE_PENALTY_OP_IMPLEMENT_STAGE2="1.0"
# OPENAI_PRESENCE_PENALTY_OP_IMPLEMENT_STAGE3="1.0"
# OPENAI_PRESENCE_PENALTY_OP_IMPLEMENT_STAGE4="1.0"
# OPENAI_PRESENCE_PENALTY_OP_DOC_STAGE1="1.0"
# OPENAI_PRESENCE_PENALTY_OP_DOC_STAGE2="1.0"
# OPENAI_PRESENCE_PENALTY_OP_EXPLAIN_STAGE1="1.0"
# OPENAI_PRESENCE_PENALTY_OP_EXPLAIN_STAGE2="1.0"
# OPENAI_PRESENCE_PENALTY="1.0"
# Options for Ollama integration, running locally.
# When using a large enough model it can produce good results for some operations. See docs/ollama.md for more info
# OLLAMA_BASE_URL="http://127.0.0.1:11434"
# Optional authentication, for use with external https proxy
# OLLAMA_AUTH_TYPE="Bearer" # Type of the authentication used, "Bearer" - api key or token (default), "Basic" - web auth with login and password.
# OLLAMA_AUTH="<your api-key or token goes here>" # When using bearer auth type, put your api key or auth token here
# OLLAMA_AUTH="<login>:<password>" # Web auth requres login and password separated by a colon
OLLAMA_MODEL_OP_ANNOTATE="qwen2.5-coder:7b-instruct-q5_K_M"
OLLAMA_MODEL_OP_ANNOTATE_POST="qwen2.5-coder:7b-instruct-q5_K_M" # used to process multiple response-variants if any
# OLLAMA_MODEL_OP_IMPLEMENT_STAGE1="qwen2.5-coder:14b-instruct-q4_K_M"
# OLLAMA_MODEL_OP_IMPLEMENT_STAGE2="qwen2.5-coder:14b-instruct-q4_K_M"
# OLLAMA_MODEL_OP_IMPLEMENT_STAGE3="qwen2.5-coder:14b-instruct-q4_K_M"
# OLLAMA_MODEL_OP_IMPLEMENT_STAGE4="qwen2.5-coder:14b-instruct-q4_K_M"
# OLLAMA_MODEL_OP_DOC_STAGE1="llama3.1:70b-instruct-q4_K_S"
# OLLAMA_MODEL_OP_DOC_STAGE2="llama3.1:70b-instruct-q4_K_S"
# OLLAMA_MODEL_OP_EXPLAIN_STAGE1="llama3.1:70b-instruct-q4_K_S"
# OLLAMA_MODEL_OP_EXPLAIN_STAGE2="llama3.1:70b-instruct-q4_K_S"
OLLAMA_MODEL="qwen2.5-coder:14b-instruct-q4_K_M"
OLLAMA_VARIANT_COUNT_OP_ANNOTATE="1" # how much annotate-response variants to generate
OLLAMA_VARIANT_SELECTION_OP_ANNOTATE="short" # how to select final variant: short, long, combine, best
OLLAMA_VARIANT_COUNT="1" # will be used as fallback
OLLAMA_VARIANT_SELECTION="short" # will be used as fallback
# Context overflow detection and management options: multiplier to increase context size, upper limit, and estimation multiplier.
# Used to detect overflow and increase context size on errors if needed. Context size will reset back to initial value when starting new operation stage.
# These params may be removed in future when Ollama implement API calls for tokenizer or prompt size statistics
OLLAMA_CONTEXT_MULT="1.75"
OLLAMA_CONTEXT_SIZE_LIMIT="65536" # reasonable limit for typical desktop systems
OLLAMA_CONTEXT_ESTIMATE_MULT="0.3"
# Context window sizes for different operations, if set too low, it will be extended automatically when context overflow detected
# If not set - use default for ollama model, and also disable context overflow detection above
OLLAMA_CONTEXT_SIZE_OP_ANNOTATE="8192"
OLLAMA_CONTEXT_SIZE_OP_ANNOTATE_POST="8192"
OLLAMA_CONTEXT_SIZE_OP_IMPLEMENT_STAGE1="24576"
OLLAMA_CONTEXT_SIZE_OP_IMPLEMENT_STAGE2="24576"
OLLAMA_CONTEXT_SIZE_OP_IMPLEMENT_STAGE3="24576"
OLLAMA_CONTEXT_SIZE_OP_IMPLEMENT_STAGE4="24576"
OLLAMA_CONTEXT_SIZE_OP_DOC_STAGE1="49152"
OLLAMA_CONTEXT_SIZE_OP_DOC_STAGE2="49152"
OLLAMA_CONTEXT_SIZE_OP_EXPLAIN_STAGE1="24576"
OLLAMA_CONTEXT_SIZE_OP_EXPLAIN_STAGE2="24576"
OLLAMA_CONTEXT_SIZE="24576"
# Switch to use structured JSON output format for some operations, may work better with some models (or not work at all)
# Supported values: plain, json. Default: plain
OLLAMA_FORMAT_OP_IMPLEMENT_STAGE1="json"
OLLAMA_FORMAT_OP_IMPLEMENT_STAGE3="json"
OLLAMA_FORMAT_OP_DOC_STAGE1="json"
OLLAMA_FORMAT_OP_EXPLAIN_STAGE1="json"
OLLAMA_MAX_TOKENS_OP_ANNOTATE="768"
OLLAMA_MAX_TOKENS_OP_ANNOTATE_POST="768"
OLLAMA_MAX_TOKENS_OP_IMPLEMENT_STAGE1="512" # file-list for review, long list is probably an error
OLLAMA_MAX_TOKENS_OP_IMPLEMENT_STAGE2="1536" # work plan also should not be too big
OLLAMA_MAX_TOKENS_OP_IMPLEMENT_STAGE3="512" # file-list for processing, long list is probably an error
OLLAMA_MAX_TOKENS_OP_IMPLEMENT_STAGE4="4096" # generated code output limit should be as big as possible
OLLAMA_MAX_TOKENS_OP_DOC_STAGE1="768" # file-list for review, long list is probably an error
OLLAMA_MAX_TOKENS_OP_DOC_STAGE2="4096" # generated document output limit should be as big as possible
OLLAMA_MAX_TOKENS_OP_EXPLAIN_STAGE1="512" # file-list for review
OLLAMA_MAX_TOKENS_OP_EXPLAIN_STAGE2="8192" # generated answer output limit
OLLAMA_MAX_TOKENS="4096"
OLLAMA_MAX_TOKENS_SEGMENTS="3"
OLLAMA_ON_FAIL_RETRIES_OP_ANNOTATE="5" # this number include errors caused by context overflow
# OLLAMA_ON_FAIL_RETRIES_OP_IMPLEMENT_STAGE1="3"
# OLLAMA_ON_FAIL_RETRIES_OP_IMPLEMENT_STAGE2="3"
# OLLAMA_ON_FAIL_RETRIES_OP_IMPLEMENT_STAGE3="3"
# OLLAMA_ON_FAIL_RETRIES_OP_IMPLEMENT_STAGE4="3"
# OLLAMA_ON_FAIL_RETRIES_OP_DOC_STAGE1="3"
# OLLAMA_ON_FAIL_RETRIES_OP_DOC_STAGE2="3"
# OLLAMA_ON_FAIL_RETRIES_OP_EXPLAIN_STAGE1="3"
# OLLAMA_ON_FAIL_RETRIES_OP_EXPLAIN_STAGE2="3"
OLLAMA_ON_FAIL_RETRIES="3"
# note: temperature highly depends on model, 0 produces mostly deterministic results
OLLAMA_TEMPERATURE_OP_ANNOTATE="0.5"
OLLAMA_TEMPERATURE_OP_ANNOTATE_POST="0"
OLLAMA_TEMPERATURE_OP_IMPLEMENT_STAGE1="0.2" # less creative for file-list output
# OLLAMA_TEMPERATURE_OP_IMPLEMENT_STAGE2="0.5"
OLLAMA_TEMPERATURE_OP_IMPLEMENT_STAGE3="0.2" # less creative for file-list output
# OLLAMA_TEMPERATURE_OP_IMPLEMENT_STAGE4="0.5"
OLLAMA_TEMPERATURE_OP_DOC_STAGE1="0.2" # less creative for file-list output
OLLAMA_TEMPERATURE_OP_DOC_STAGE2="0.7"
OLLAMA_TEMPERATURE_OP_EXPLAIN_STAGE1="0.2" # less creative for file-list output
OLLAMA_TEMPERATURE_OP_EXPLAIN_STAGE2="0.7"
OLLAMA_TEMPERATURE="0.5"
# System prompt role for model, can be configured per operation. Useful if model not supporting system prompt
# Valid values: system, user. default: system.
# When using "user" role, system prompt will be converted to user-query + ai-acknowledge message sequence
# OLLAMA_SYSPROMPT_ROLE_OP_ANNOTATE="system"
# OLLAMA_SYSPROMPT_ROLE_OP_ANNOTATE_POST="system"
# OLLAMA_SYSPROMPT_ROLE_OP_IMPLEMENT_STAGE1="system"
# OLLAMA_SYSPROMPT_ROLE_OP_IMPLEMENT_STAGE2="system"
# OLLAMA_SYSPROMPT_ROLE_OP_IMPLEMENT_STAGE3="system"
# OLLAMA_SYSPROMPT_ROLE_OP_IMPLEMENT_STAGE4="system"
# OLLAMA_SYSPROMPT_ROLE_OP_DOC_STAGE1="system"
# OLLAMA_SYSPROMPT_ROLE_OP_DOC_STAGE2="system"
OLLAMA_SYSPROMPT_ROLE_OP_EXPLAIN_STAGE1="system"
OLLAMA_SYSPROMPT_ROLE_OP_EXPLAIN_STAGE2="system"
# OLLAMA_SYSPROMPT_ROLE="system"
# Optional regexps for filtering out responses from reasoning models, like deepseek r1
# THINK-regexps will be used to remove reasoning part from response L - is for opening tag, R - is for closing tag
# OUT-regexps will be used to extract output part from response after it was filered out with THINK-regexps
# OLLAMA_THINK_RX_L_OP_ANNOTATE="(?mi)^\\s*<think>\\s*\$"
# OLLAMA_THINK_RX_R_OP_ANNOTATE="(?mi)^\\s*</think>\\s*\$"
# OLLAMA_OUT_RX_L_OP_ANNOTATE="(?mi)^\\s*<output>\\s*\$"
# OLLAMA_OUT_RX_R_OP_ANNOTATE="(?mi)^\\s*</output>\\s*\$"
# OLLAMA_THINK_RX_L_OP_ANNOTATE_POST="(?mi)^\\s*<think>\\s*\$"
# OLLAMA_THINK_RX_R_OP_ANNOTATE_POST="(?mi)^\\s*</think>\\s*\$"
# OLLAMA_OUT_RX_L_OP_ANNOTATE_POST="(?mi)^\\s*<output>\\s*\$"
# OLLAMA_OUT_RX_R_OP_ANNOTATE_POST="(?mi)^\\s*</output>\\s*\$"
# OLLAMA_THINK_RX_L_OP_IMPLEMENT_STAGE1="(?mi)^\\s*<think>\\s*\$"
# OLLAMA_THINK_RX_R_OP_IMPLEMENT_STAGE1="(?mi)^\\s*</think>\\s*\$"
# OLLAMA_OUT_RX_L_OP_IMPLEMENT_STAGE1="(?mi)^\\s*<output>\\s*\$"
# OLLAMA_OUT_RX_R_OP_IMPLEMENT_STAGE1="(?mi)^\\s*</output>\\s*\$"
# OLLAMA_THINK_RX_L_OP_IMPLEMENT_STAGE2="(?mi)^\\s*<think>\\s*\$"
# OLLAMA_THINK_RX_R_OP_IMPLEMENT_STAGE2="(?mi)^\\s*</think>\\s*\$"
# OLLAMA_OUT_RX_L_OP_IMPLEMENT_STAGE2="(?mi)^\\s*<output>\\s*\$"
# OLLAMA_OUT_RX_R_OP_IMPLEMENT_STAGE2="(?mi)^\\s*</output>\\s*\$"
# OLLAMA_THINK_RX_L_OP_IMPLEMENT_STAGE3="(?mi)^\\s*<think>\\s*\$"
# OLLAMA_THINK_RX_R_OP_IMPLEMENT_STAGE3="(?mi)^\\s*</think>\\s*\$"
# OLLAMA_OUT_RX_L_OP_IMPLEMENT_STAGE3="(?mi)^\\s*<output>\\s*\$"
# OLLAMA_OUT_RX_R_OP_IMPLEMENT_STAGE3="(?mi)^\\s*</output>\\s*\$"
# OLLAMA_THINK_RX_L_OP_IMPLEMENT_STAGE4="(?mi)^\\s*<think>\\s*\$"
# OLLAMA_THINK_RX_R_OP_IMPLEMENT_STAGE4="(?mi)^\\s*</think>\\s*\$"
# OLLAMA_OUT_RX_L_OP_IMPLEMENT_STAGE4="(?mi)^\\s*<output>\\s*\$"
# OLLAMA_OUT_RX_R_OP_IMPLEMENT_STAGE4="(?mi)^\\s*</output>\\s*\$"
# OLLAMA_THINK_RX_L_OP_DOC_STAGE1="(?mi)^\\s*<think>\\s*\$"
# OLLAMA_THINK_RX_R_OP_DOC_STAGE1="(?mi)^\\s*</think>\\s*\$"
# OLLAMA_OUT_RX_L_OP_DOC_STAGE1="(?mi)^\\s*<output>\\s*\$"
# OLLAMA_OUT_RX_R_OP_DOC_STAGE1="(?mi)^\\s*</output>\\s*\$"
# OLLAMA_THINK_RX_L_OP_DOC_STAGE2="(?mi)^\\s*<think>\\s*\$"
# OLLAMA_THINK_RX_R_OP_DOC_STAGE2="(?mi)^\\s*</think>\\s*\$"
# OLLAMA_OUT_RX_L_OP_DOC_STAGE2="(?mi)^\\s*<output>\\s*\$"
# OLLAMA_OUT_RX_R_OP_DOC_STAGE2="(?mi)^\\s*</output>\\s*\$"
# OLLAMA_THINK_RX_L_OP_EXPLAIN_STAGE1="(?mi)^\\s*<think>\\s*\$"
# OLLAMA_THINK_RX_R_OP_EXPLAIN_STAGE1="(?mi)^\\s*</think>\\s*\$"
# OLLAMA_OUT_RX_L_OP_EXPLAIN_STAGE1="(?mi)^\\s*<output>\\s*\$"
# OLLAMA_OUT_RX_R_OP_EXPLAIN_STAGE1="(?mi)^\\s*</output>\\s*\$"
# OLLAMA_THINK_RX_L_OP_EXPLAIN_STAGE2="(?mi)^\\s*<think>\\s*\$"
# OLLAMA_THINK_RX_R_OP_EXPLAIN_STAGE2="(?mi)^\\s*</think>\\s*\$"
# OLLAMA_OUT_RX_L_OP_EXPLAIN_STAGE2="(?mi)^\\s*<output>\\s*\$"
# OLLAMA_OUT_RX_R_OP_EXPLAIN_STAGE2="(?mi)^\\s*</output>\\s*\$"
# OLLAMA_THINK_RX_L="(?mi)^\\s*<think>\\s*\$"
# OLLAMA_THINK_RX_R="(?mi)^\\s*</think>\\s*\$"
# OLLAMA_OUT_RX_L="(?mi)^\\s*<output>\\s*\$"
# OLLAMA_OUT_RX_R="(?mi)^\\s*</output>\\s*\$"
# The remaining options are for fine-tuning the model, when using with smaller sub-15b models, their use may be cruical to make things work
# OLLAMA_TOP_K_OP_ANNOTATE="40"
# OLLAMA_TOP_K_OP_ANNOTATE_POST="40"
# OLLAMA_TOP_K_OP_IMPLEMENT_STAGE1="40"
# OLLAMA_TOP_K_OP_IMPLEMENT_STAGE2="40"
# OLLAMA_TOP_K_OP_IMPLEMENT_STAGE3="40"
# OLLAMA_TOP_K_OP_IMPLEMENT_STAGE4="40"
# OLLAMA_TOP_K_OP_DOC_STAGE1="40"
# OLLAMA_TOP_K_OP_DOC_STAGE2="40"
# OLLAMA_TOP_K_OP_EXPLAIN_STAGE1="40"
# OLLAMA_TOP_K_OP_EXPLAIN_STAGE2="40"
# OLLAMA_TOP_K="40"
# OLLAMA_TOP_P_OP_ANNOTATE="0.9"
# OLLAMA_TOP_P_OP_ANNOTATE_POST="0.9"
# OLLAMA_TOP_P_OP_IMPLEMENT_STAGE1="0.9"
# OLLAMA_TOP_P_OP_IMPLEMENT_STAGE2="0.9"
# OLLAMA_TOP_P_OP_IMPLEMENT_STAGE3="0.9"
# OLLAMA_TOP_P_OP_IMPLEMENT_STAGE4="0.9"
# OLLAMA_TOP_P_OP_DOC_STAGE1="0.9"
# OLLAMA_TOP_P_OP_DOC_STAGE2="0.9"
# OLLAMA_TOP_P_OP_EXPLAIN_STAGE1="0.9"
# OLLAMA_TOP_P_OP_EXPLAIN_STAGE2="0.9"
# OLLAMA_TOP_P="0.9"
# OLLAMA_SEED_OP_ANNOTATE="42"
# OLLAMA_SEED_OP_ANNOTATE_POST="42"
# OLLAMA_SEED_OP_IMPLEMENT_STAGE1="42"
# OLLAMA_SEED_OP_IMPLEMENT_STAGE2="42"
# OLLAMA_SEED_OP_IMPLEMENT_STAGE3="42"
# OLLAMA_SEED_OP_IMPLEMENT_STAGE4="42"
# OLLAMA_SEED_OP_DOC_STAGE1="42"
# OLLAMA_SEED_OP_DOC_STAGE2="42"
# OLLAMA_SEED_OP_EXPLAIN_STAGE1="42"
# OLLAMA_SEED_OP_EXPLAIN_STAGE2="42"
# OLLAMA_SEED="42"
# note: values slightly more than 1.0 seem to help against problems when LLM starts to generate repeated content indefinitely, without making report to omit important items
OLLAMA_REPEAT_PENALTY_OP_ANNOTATE="1.1"
OLLAMA_REPEAT_PENALTY_OP_ANNOTATE_POST="1.1"
OLLAMA_REPEAT_PENALTY_OP_IMPLEMENT_STAGE1="1.0"
OLLAMA_REPEAT_PENALTY_OP_IMPLEMENT_STAGE2="1.1"
OLLAMA_REPEAT_PENALTY_OP_IMPLEMENT_STAGE3="1.0"
OLLAMA_REPEAT_PENALTY_OP_IMPLEMENT_STAGE4="1.0"
# OLLAMA_REPEAT_PENALTY_OP_DOC_STAGE1="1.2"
# OLLAMA_REPEAT_PENALTY_OP_DOC_STAGE2="1.2"
OLLAMA_REPEAT_PENALTY_OP_EXPLAIN_STAGE1="1.0"
OLLAMA_REPEAT_PENALTY_OP_EXPLAIN_STAGE2="1.1"
# OLLAMA_REPEAT_PENALTY="1.0"
# OLLAMA_FREQ_PENALTY_OP_ANNOTATE="1.0"
# OLLAMA_FREQ_PENALTY_OP_ANNOTATE_POST="1.0"
# OLLAMA_FREQ_PENALTY_OP_IMPLEMENT_STAGE1="1.0"
# OLLAMA_FREQ_PENALTY_OP_IMPLEMENT_STAGE2="1.0"
# OLLAMA_FREQ_PENALTY_OP_IMPLEMENT_STAGE3="1.0"
# OLLAMA_FREQ_PENALTY_OP_IMPLEMENT_STAGE4="1.0"
# OLLAMA_FREQ_PENALTY_OP_DOC_STAGE1="1.0"
# OLLAMA_FREQ_PENALTY_OP_DOC_STAGE2="1.0"
# OLLAMA_FREQ_PENALTY_OP_EXPLAIN_STAGE1="1.0"
# OLLAMA_FREQ_PENALTY_OP_EXPLAIN_STAGE2="1.0"
# OLLAMA_FREQ_PENALTY="1.0"
# OLLAMA_PRESENCE_PENALTY_OP_ANNOTATE="1.0"
# OLLAMA_PRESENCE_PENALTY_OP_ANNOTATE_POST="1.0"
# OLLAMA_PRESENCE_PENALTY_OP_IMPLEMENT_STAGE1="1.0"
# OLLAMA_PRESENCE_PENALTY_OP_IMPLEMENT_STAGE2="1.0"
# OLLAMA_PRESENCE_PENALTY_OP_IMPLEMENT_STAGE3="1.0"
# OLLAMA_PRESENCE_PENALTY_OP_IMPLEMENT_STAGE4="1.0"
# OLLAMA_PRESENCE_PENALTY_OP_DOC_STAGE1="1.0"
# OLLAMA_PRESENCE_PENALTY_OP_DOC_STAGE2="1.0"
# OLLAMA_PRESENCE_PENALTY_OP_EXPLAIN_STAGE1="1.0"
# OLLAMA_PRESENCE_PENALTY_OP_EXPLAIN_STAGE2="1.0"
# OLLAMA_PRESENCE_PENALTY="1.0"
# Options for generic provider with OpenAI compatible API, JSON structured output mode is not supported
# Below is example for deepseek (https://www.deepseek.com/)
GENERIC_BASE_URL="https://api.deepseek.com/v1" # Required parameter for generic provider (example for deepseek)
# Authentication, optional, but most likely required by your provider, GENERIC_API_KEY="<token>" is now deprecated, will work same as GENERIC_AUTH
# GENERIC_AUTH_TYPE="Bearer" # Type of the authentication used, "Bearer" - api key or token (default), "Basic" - web auth with login and password.
# GENERIC_AUTH="<your api-key or token goes here>" # When using bearer auth type, put your api key or auth token here
# GENERIC_AUTH="<login>:<password>" # Web auth requres login and password separated by a colon
# Some system parameters, provider dependent
# GENERIC_ENABLE_STREAMING="0" # 0 - disabled (default), 1 - enabled, write new data to log right after it generated by LLM, useful for debugging
# GENERIC_MAXTOKENS_FORMAT="old" # values: old, new. Old is "max_tokens=<value>" (default), New is "max_completion_tokens=<value>" (OpenAI using it now)
# General parameters for different operations
# GENERIC_MODEL_OP_ANNOTATE="deepseek-chat"
# GENERIC_MODEL_OP_ANNOTATE_POST="deepseek-chat" # used to process multiple response-variants if any
# GENERIC_MODEL_OP_IMPLEMENT_STAGE1="deepseek-chat"
# GENERIC_MODEL_OP_IMPLEMENT_STAGE2="deepseek-chat"
# GENERIC_MODEL_OP_IMPLEMENT_STAGE3="deepseek-chat"
# GENERIC_MODEL_OP_IMPLEMENT_STAGE4="deepseek-chat"
# GENERIC_MODEL_OP_DOC_STAGE1="deepseek-chat"
# GENERIC_MODEL_OP_DOC_STAGE2="deepseek-chat"
# GENERIC_MODEL_OP_EXPLAIN_STAGE1="deepseek-chat"
# GENERIC_MODEL_OP_EXPLAIN_STAGE2="deepseek-chat"
GENERIC_MODEL="deepseek-chat"
# GENERIC_MODEL="deepseek-reasoner" # do not forget to set system prompt role to "user" below
GENERIC_VARIANT_COUNT_OP_ANNOTATE="1" # how much annotate-response variants to generate
GENERIC_VARIANT_SELECTION_OP_ANNOTATE="short" # how to select final variant: short, long, combine, best
GENERIC_VARIANT_COUNT="1" # will be used as fallback
GENERIC_VARIANT_SELECTION="short" # will be used as fallback
GENERIC_MAX_TOKENS_OP_ANNOTATE="768"
GENERIC_MAX_TOKENS_OP_ANNOTATE_POST="768"
GENERIC_MAX_TOKENS_OP_IMPLEMENT_STAGE1="512" # file-list for review, long list is probably an error
GENERIC_MAX_TOKENS_OP_IMPLEMENT_STAGE2="1536" # work plan also should not be too big
GENERIC_MAX_TOKENS_OP_IMPLEMENT_STAGE3="512" # file-list for processing, long list is probably an error
GENERIC_MAX_TOKENS_OP_IMPLEMENT_STAGE4="8192" # generated code output limit should be as big as possible, works with DeepSeek
GENERIC_MAX_TOKENS_OP_DOC_STAGE1="768" # file-list for review, longer list is probably an error
GENERIC_MAX_TOKENS_OP_DOC_STAGE2="8192" # generated document output limit should be as big as possible, works with DeepSeek
GENERIC_MAX_TOKENS_OP_EXPLAIN_STAGE1="512" # file-list for review
GENERIC_MAX_TOKENS_OP_EXPLAIN_STAGE2="8192" # generated answer output limit
GENERIC_MAX_TOKENS="4096" # probably should work with most LLMs
GENERIC_MAX_TOKENS_SEGMENTS="3"
GENERIC_ON_FAIL_RETRIES_OP_ANNOTATE="1"
# GENERIC_ON_FAIL_RETRIES_OP_IMPLEMENT_STAGE1="3"
# GENERIC_ON_FAIL_RETRIES_OP_IMPLEMENT_STAGE2="3"
# GENERIC_ON_FAIL_RETRIES_OP_IMPLEMENT_STAGE3="3"
# GENERIC_ON_FAIL_RETRIES_OP_IMPLEMENT_STAGE4="3"
# GENERIC_ON_FAIL_RETRIES_OP_DOC_STAGE1="3"
# GENERIC_ON_FAIL_RETRIES_OP_DOC_STAGE2="3"
# GENERIC_ON_FAIL_RETRIES_OP_EXPLAIN_STAGE1="3"
# GENERIC_ON_FAIL_RETRIES_OP_EXPLAIN_STAGE2="3"
GENERIC_ON_FAIL_RETRIES="3"
# Temperature typically affecting LLM behavior the most and it provider dependent, refer your provider's API docs for value limits
# Values provided below seem to work good with DeepSeek
# GENERIC_TEMPERATURE_OP_ANNOTATE="0.5"
# GENERIC_TEMPERATURE_OP_ANNOTATE_POST="0.5"
GENERIC_TEMPERATURE_OP_IMPLEMENT_STAGE1="0.2" # less creative for file-list output
# GENERIC_TEMPERATURE_OP_IMPLEMENT_STAGE2="0.5"
GENERIC_TEMPERATURE_OP_IMPLEMENT_STAGE3="0.2" # less creative for file-list output
# GENERIC_TEMPERATURE_OP_IMPLEMENT_STAGE4="0.5"
GENERIC_TEMPERATURE_OP_DOC_STAGE1="0.2" # less creative for file-list output
GENERIC_TEMPERATURE_OP_DOC_STAGE2="0.7" # more creative when writing documentation
GENERIC_TEMPERATURE_OP_EXPLAIN_STAGE1="0.2" # less creative for file-list output
GENERIC_TEMPERATURE_OP_EXPLAIN_STAGE2="0.7"
GENERIC_TEMPERATURE="0.5"
# System prompt role for model, can be configured per operation. Useful for reasoning models without system prompt, like deepseek-reasoner
# Valid values: system, developer, user. default: system.
# When using "user" role, system prompt will be converted to user-query + ai-acknowledge message sequence
# GENERIC_SYSPROMPT_ROLE_OP_ANNOTATE="system"
# GENERIC_SYSPROMPT_ROLE_OP_ANNOTATE_POST="system"
# GENERIC_SYSPROMPT_ROLE_OP_IMPLEMENT_STAGE1="system"
# GENERIC_SYSPROMPT_ROLE_OP_IMPLEMENT_STAGE2="system"
# GENERIC_SYSPROMPT_ROLE_OP_IMPLEMENT_STAGE3="system"
# GENERIC_SYSPROMPT_ROLE_OP_IMPLEMENT_STAGE4="system"
# GENERIC_SYSPROMPT_ROLE_OP_DOC_STAGE1="system"
# GENERIC_SYSPROMPT_ROLE_OP_DOC_STAGE2="system"
# GENERIC_SYSPROMPT_ROLE_OP_EXPLAIN_STAGE1="system"
# GENERIC_SYSPROMPT_ROLE_OP_EXPLAIN_STAGE2="system"
# GENERIC_SYSPROMPT_ROLE="system"
# Optional regexps for filtering out responses from reasoning models, if LLM provider not doing this automatically
# THINK-regexps will be used to remove reasoning part from response L - is for opening tag, R - is for closing tag
# OUT-regexps will be used to extract output part from response after it was filered out with THINK-regexps
# GENERIC_THINK_RX_L_OP_ANNOTATE="(?mi)^\\s*<think>\\s*\$"
# GENERIC_THINK_RX_R_OP_ANNOTATE="(?mi)^\\s*</think>\\s*\$"
# GENERIC_OUT_RX_L_OP_ANNOTATE="(?mi)^\\s*<output>\\s*\$"
# GENERIC_OUT_RX_R_OP_ANNOTATE="(?mi)^\\s*</output>\\s*\$"
# GENERIC_THINK_RX_L_OP_ANNOTATE_POST="(?mi)^\\s*<think>\\s*\$"
# GENERIC_THINK_RX_R_OP_ANNOTATE_POST="(?mi)^\\s*</think>\\s*\$"
# GENERIC_OUT_RX_L_OP_ANNOTATE_POST="(?mi)^\\s*<output>\\s*\$"
# GENERIC_OUT_RX_R_OP_ANNOTATE_POST="(?mi)^\\s*</output>\\s*\$"
# GENERIC_THINK_RX_L_OP_IMPLEMENT_STAGE1="(?mi)^\\s*<think>\\s*\$"
# GENERIC_THINK_RX_R_OP_IMPLEMENT_STAGE1="(?mi)^\\s*</think>\\s*\$"
# GENERIC_OUT_RX_L_OP_IMPLEMENT_STAGE1="(?mi)^\\s*<output>\\s*\$"
# GENERIC_OUT_RX_R_OP_IMPLEMENT_STAGE1="(?mi)^\\s*</output>\\s*\$"
# GENERIC_THINK_RX_L_OP_IMPLEMENT_STAGE2="(?mi)^\\s*<think>\\s*\$"
# GENERIC_THINK_RX_R_OP_IMPLEMENT_STAGE2="(?mi)^\\s*</think>\\s*\$"
# GENERIC_OUT_RX_L_OP_IMPLEMENT_STAGE2="(?mi)^\\s*<output>\\s*\$"
# GENERIC_OUT_RX_R_OP_IMPLEMENT_STAGE2="(?mi)^\\s*</output>\\s*\$"
# GENERIC_THINK_RX_L_OP_IMPLEMENT_STAGE3="(?mi)^\\s*<think>\\s*\$"
# GENERIC_THINK_RX_R_OP_IMPLEMENT_STAGE3="(?mi)^\\s*</think>\\s*\$"
# GENERIC_OUT_RX_L_OP_IMPLEMENT_STAGE3="(?mi)^\\s*<output>\\s*\$"
# GENERIC_OUT_RX_R_OP_IMPLEMENT_STAGE3="(?mi)^\\s*</output>\\s*\$"
# GENERIC_THINK_RX_L_OP_IMPLEMENT_STAGE4="(?mi)^\\s*<think>\\s*\$"
# GENERIC_THINK_RX_R_OP_IMPLEMENT_STAGE4="(?mi)^\\s*</think>\\s*\$"
# GENERIC_OUT_RX_L_OP_IMPLEMENT_STAGE4="(?mi)^\\s*<output>\\s*\$"
# GENERIC_OUT_RX_R_OP_IMPLEMENT_STAGE4="(?mi)^\\s*</output>\\s*\$"
# GENERIC_THINK_RX_L_OP_DOC_STAGE1="(?mi)^\\s*<think>\\s*\$"
# GENERIC_THINK_RX_R_OP_DOC_STAGE1="(?mi)^\\s*</think>\\s*\$"
# GENERIC_OUT_RX_L_OP_DOC_STAGE1="(?mi)^\\s*<output>\\s*\$"
# GENERIC_OUT_RX_R_OP_DOC_STAGE1="(?mi)^\\s*</output>\\s*\$"
# GENERIC_THINK_RX_L_OP_DOC_STAGE2="(?mi)^\\s*<think>\\s*\$"
# GENERIC_THINK_RX_R_OP_DOC_STAGE2="(?mi)^\\s*</think>\\s*\$"
# GENERIC_OUT_RX_L_OP_DOC_STAGE2="(?mi)^\\s*<output>\\s*\$"
# GENERIC_OUT_RX_R_OP_DOC_STAGE2="(?mi)^\\s*</output>\\s*\$"
# GENERIC_THINK_RX_L_OP_EXPLAIN_STAGE1="(?mi)^\\s*<think>\\s*\$"
# GENERIC_THINK_RX_R_OP_EXPLAIN_STAGE1="(?mi)^\\s*</think>\\s*\$"
# GENERIC_OUT_RX_L_OP_EXPLAIN_STAGE1="(?mi)^\\s*<output>\\s*\$"
# GENERIC_OUT_RX_R_OP_EXPLAIN_STAGE1="(?mi)^\\s*</output>\\s*\$"
# GENERIC_THINK_RX_L_OP_EXPLAIN_STAGE2="(?mi)^\\s*<think>\\s*\$"
# GENERIC_THINK_RX_R_OP_EXPLAIN_STAGE2="(?mi)^\\s*</think>\\s*\$"
# GENERIC_OUT_RX_L_OP_EXPLAIN_STAGE2="(?mi)^\\s*<output>\\s*\$"
# GENERIC_OUT_RX_R_OP_EXPLAIN_STAGE2="(?mi)^\\s*</output>\\s*\$"
# GENERIC_THINK_RX_L="(?mi)^\\s*<think>\\s*\$"
# GENERIC_THINK_RX_R="(?mi)^\\s*</think>\\s*\$"
# GENERIC_OUT_RX_L="(?mi)^\\s*<output>\\s*\$"
# GENERIC_OUT_RX_R="(?mi)^\\s*</output>\\s*\$"
# Remaining options may or may not work, depending on LLM provider
# GENERIC_TOP_K_OP_ANNOTATE="40"
# GENERIC_TOP_K_OP_ANNOTATE_POST="40"
# GENERIC_TOP_K_OP_IMPLEMENT_STAGE1="40"
# GENERIC_TOP_K_OP_IMPLEMENT_STAGE2="40"
# GENERIC_TOP_K_OP_IMPLEMENT_STAGE3="40"
# GENERIC_TOP_K_OP_IMPLEMENT_STAGE4="40"
# GENERIC_TOP_K_OP_DOC_STAGE1="40"
# GENERIC_TOP_K_OP_DOC_STAGE2="40"
# GENERIC_TOP_K_OP_EXPLAIN_STAGE1="40"
# GENERIC_TOP_K_OP_EXPLAIN_STAGE2="40"
# GENERIC_TOP_K="40"
# GENERIC_TOP_P_OP_ANNOTATE="0.9"
# GENERIC_TOP_P_OP_ANNOTATE_POST="0.9"
# GENERIC_TOP_P_OP_IMPLEMENT_STAGE1="0.9"
# GENERIC_TOP_P_OP_IMPLEMENT_STAGE2="0.9"
# GENERIC_TOP_P_OP_IMPLEMENT_STAGE3="0.9"
# GENERIC_TOP_P_OP_IMPLEMENT_STAGE4="0.9"
# GENERIC_TOP_P_OP_DOC_STAGE1="0.9"
# GENERIC_TOP_P_OP_DOC_STAGE2="0.9"
# GENERIC_TOP_P_OP_EXPLAIN_STAGE1="0.9"
# GENERIC_TOP_P_OP_EXPLAIN_STAGE2="0.9"
# GENERIC_TOP_P="0.9"
# GENERIC_SEED_OP_ANNOTATE="42"
# GENERIC_SEED_OP_ANNOTATE_POST="42"
# GENERIC_SEED_OP_IMPLEMENT_STAGE1="42"
# GENERIC_SEED_OP_IMPLEMENT_STAGE2="42"
# GENERIC_SEED_OP_IMPLEMENT_STAGE3="42"
# GENERIC_SEED_OP_IMPLEMENT_STAGE4="42"
# GENERIC_SEED_OP_DOC_STAGE1="42"
# GENERIC_SEED_OP_DOC_STAGE2="42"
# GENERIC_SEED_OP_EXPLAIN_STAGE1="42"
# GENERIC_SEED_OP_EXPLAIN_STAGE2="42"
# GENERIC_SEED="42"
# GENERIC_REPEAT_PENALTY_OP_ANNOTATE="1.2"
# GENERIC_REPEAT_PENALTY_OP_ANNOTATE_POST="1.2"
# GENERIC_REPEAT_PENALTY_OP_IMPLEMENT_STAGE1="1.2"
# GENERIC_REPEAT_PENALTY_OP_IMPLEMENT_STAGE2="1.5"
# GENERIC_REPEAT_PENALTY_OP_IMPLEMENT_STAGE3="1.0"
# GENERIC_REPEAT_PENALTY_OP_IMPLEMENT_STAGE4="1.0"
# GENERIC_REPEAT_PENALTY_OP_DOC_STAGE1="1.2"
# GENERIC_REPEAT_PENALTY_OP_DOC_STAGE2="1.2"
# GENERIC_REPEAT_PENALTY_OP_EXPLAIN_STAGE1="1.0"
# GENERIC_REPEAT_PENALTY_OP_EXPLAIN_STAGE2="1.1"
# GENERIC_REPEAT_PENALTY="1.1"
# GENERIC_FREQ_PENALTY_OP_ANNOTATE="1.0"
# GENERIC_FREQ_PENALTY_OP_ANNOTATE_POST="1.0"
# GENERIC_FREQ_PENALTY_OP_IMPLEMENT_STAGE1="1.0"
# GENERIC_FREQ_PENALTY_OP_IMPLEMENT_STAGE2="1.0"
# GENERIC_FREQ_PENALTY_OP_IMPLEMENT_STAGE3="1.0"
# GENERIC_FREQ_PENALTY_OP_IMPLEMENT_STAGE4="1.0"
# GENERIC_FREQ_PENALTY_OP_DOC_STAGE1="1.0"
# GENERIC_FREQ_PENALTY_OP_DOC_STAGE2="1.0"
# GENERIC_FREQ_PENALTY_OP_EXPLAIN_STAGE1="1.0"
# GENERIC_FREQ_PENALTY_OP_EXPLAIN_STAGE2="1.0"
# GENERIC_FREQ_PENALTY="1.0"
# GENERIC_PRESENCE_PENALTY_OP_ANNOTATE="1.0"
# GENERIC_PRESENCE_PENALTY_OP_ANNOTATE_POST="1.0"
# GENERIC_PRESENCE_PENALTY_OP_IMPLEMENT_STAGE1="1.0"
# GENERIC_PRESENCE_PENALTY_OP_IMPLEMENT_STAGE2="1.0"
# GENERIC_PRESENCE_PENALTY_OP_IMPLEMENT_STAGE3="1.0"
# GENERIC_PRESENCE_PENALTY_OP_IMPLEMENT_STAGE4="1.0"
# GENERIC_PRESENCE_PENALTY_OP_DOC_STAGE1="1.0"
# GENERIC_PRESENCE_PENALTY_OP_DOC_STAGE2="1.0"
# GENERIC_PRESENCE_PENALTY_OP_EXPLAIN_STAGE1="1.0"
# GENERIC_PRESENCE_PENALTY_OP_EXPLAIN_STAGE2="1.0"
# GENERIC_PRESENCE_PENALTY="1.0"
# GENERIC_REASONING_EFFORT_OP_ANNOTATE="medium" # values: low, medium, high
# GENERIC_REASONING_EFFORT_OP_ANNOTATE_POST="medium"
# GENERIC_REASONING_EFFORT_OP_IMPLEMENT_STAGE1="medium"
# GENERIC_REASONING_EFFORT_OP_IMPLEMENT_STAGE2="medium"
# GENERIC_REASONING_EFFORT_OP_IMPLEMENT_STAGE3="medium"
# GENERIC_REASONING_EFFORT_OP_IMPLEMENT_STAGE4="medium"
# GENERIC_REASONING_EFFORT_OP_DOC_STAGE1="medium"
# GENERIC_REASONING_EFFORT_OP_DOC_STAGE2="medium"
# GENERIC_REASONING_EFFORT_OP_EXPLAIN_STAGE1="medium"
# GENERIC_REASONING_EFFORT_OP_EXPLAIN_STAGE2="medium"
# GENERIC_REASONING_EFFORT="medium"