@eigenpal/cli 0.20.62 → 0.20.64
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/CHANGELOG.md +13 -1
- package/dist/cli.js +21 -10
- package/dist/docs-bundle.json +27 -13
- package/dist/skill/reference/debugging.md +4 -3
- package/dist/skill/reference/evaluators.md +3 -2
- package/dist/skill/reference/step-types.md +6 -6
- package/dist/skill/reference/workflow-yaml.md +2 -2
- package/package.json +1 -1
|
@@ -131,9 +131,10 @@ eigenpal workflow step-type list --search <partial-match>
|
|
|
131
131
|
|
|
132
132
|
### Eval flakiness with `llm-judge`
|
|
133
133
|
|
|
134
|
-
LLM-judge scores have variance. Stabilize:
|
|
135
|
-
(`gpt-
|
|
136
|
-
dimension per judge, run multiple experiments and average.
|
|
134
|
+
LLM-judge scores have variance. Stabilize: select a current model
|
|
135
|
+
(`openai/gpt-6.1-sol`), keep its reasoning effort fixed, narrow the prompt to
|
|
136
|
+
one dimension per judge, run multiple experiments and average. Only set
|
|
137
|
+
temperature when the model supports it; reasoning models may reject it.
|
|
137
138
|
|
|
138
139
|
### Workflow `Validation failed` on push
|
|
139
140
|
|
|
@@ -145,7 +145,7 @@ paths; you do not need a separate evaluator for failure cases.
|
|
|
145
145
|
- name: covenants-recall
|
|
146
146
|
type: llm-judge
|
|
147
147
|
config:
|
|
148
|
-
model: gpt-
|
|
148
|
+
model: gpt-6-luna # Optional, falls back to the workspace default LLM provider
|
|
149
149
|
mode: continuous # continuous | discrete (default: continuous)
|
|
150
150
|
passThreshold: 0.85
|
|
151
151
|
docLabel: extraction # what the judge calls the workflow output (default: "output")
|
|
@@ -297,7 +297,8 @@ process — `--out` writes the file directly from the server's storage.
|
|
|
297
297
|
|
|
298
298
|
LLM-judge scores have variance. To reduce flake:
|
|
299
299
|
|
|
300
|
-
-
|
|
300
|
+
- Select a current model (`openai/gpt-6.1-sol`) and keep its reasoning effort fixed.
|
|
301
|
+
Re-run representative evaluations when changing models or when the provider updates it.
|
|
301
302
|
- Keep `promptExtension` narrow — score one dimension per evaluator. Use
|
|
302
303
|
multiple judges instead of one omnibus prompt.
|
|
303
304
|
- Prefer `mode: discrete` with a small label set when the judgement is
|
|
@@ -2808,7 +2808,7 @@ Extract structured data from text using AI with a JSON schema
|
|
|
2808
2808
|
| `input` | string | yes | | Text content or template expression |
|
|
2809
2809
|
| `schema` | object | yes | | JSON Schema defining the structure to extract |
|
|
2810
2810
|
| `prompt` | string | no | | Custom prompt template for extraction |
|
|
2811
|
-
| `provider` | string | no | | Provider ID (e.g., "openai-
|
|
2811
|
+
| `provider` | string | no | | Provider ID (e.g., "openai/gpt-6-luna") |
|
|
2812
2812
|
| `model` | string | no | | Model override |
|
|
2813
2813
|
| `reasoningEffort` | `"none"` \| `"minimal"` \| `"low"` \| `"medium"` \| `"high"` \| `"xhigh"` \| `"max"` | no | | Reasoning effort for models that support it. Omit to use the selected model's default. |
|
|
2814
2814
|
| `maxInputTokens` | integer | no | | Max input tokens. Truncates input text and logs a warning when exceeded. Omit for no limit. |
|
|
@@ -2868,7 +2868,7 @@ Config schema:
|
|
|
2868
2868
|
"type": "string"
|
|
2869
2869
|
},
|
|
2870
2870
|
"provider": {
|
|
2871
|
-
"description": "Provider ID (e.g., \"openai-
|
|
2871
|
+
"description": "Provider ID (e.g., \"openai/gpt-6-luna\")",
|
|
2872
2872
|
"type": "string"
|
|
2873
2873
|
},
|
|
2874
2874
|
"model": {
|
|
@@ -2999,7 +2999,7 @@ Split a parsed document into named sections using an LLM. Consumes ai.parse outp
|
|
|
2999
2999
|
| `input` | string | yes | | Template expression resolving to ai.parse output, e.g. "{{ steps.parse.output }}" |
|
|
3000
3000
|
| `sections` | array<object> | yes | | Named sections to find in the document |
|
|
3001
3001
|
| `rules` | string | no | | Optional natural-language rules appended to the system prompt. E.g. "End-of-section markers like *Koniec prílohy 2* close the current section." |
|
|
3002
|
-
| `provider` | string | no | | Provider ID from eigenpal.config.yaml (e.g. "openai-
|
|
3002
|
+
| `provider` | string | no | | Provider ID from eigenpal.config.yaml (e.g. "openai/gpt-6-luna"). Falls back to the tenant default LLM provider when omitted. |
|
|
3003
3003
|
| `reasoningEffort` | `"none"` \| `"minimal"` \| `"low"` \| `"medium"` \| `"high"` \| `"xhigh"` \| `"max"` | no | | Reasoning effort for models that support it. Omit to use the selected model's default. |
|
|
3004
3004
|
| `windowTokenBudget` | integer | no | | Override the estimated per-window token ceiling. Oversized pages are subdivided with overlap while retaining their original page index. Defaults to env SPLIT_WINDOW_TOKEN_BUDGET or 20000. Inputs requiring more than 64 LLM windows are rejected before calls begin. |
|
|
3005
3005
|
|
|
@@ -3064,7 +3064,7 @@ Config schema:
|
|
|
3064
3064
|
"type": "string"
|
|
3065
3065
|
},
|
|
3066
3066
|
"provider": {
|
|
3067
|
-
"description": "Provider ID from eigenpal.config.yaml (e.g. \"openai-
|
|
3067
|
+
"description": "Provider ID from eigenpal.config.yaml (e.g. \"openai/gpt-6-luna\"). Falls back to the tenant default LLM provider when omitted.",
|
|
3068
3068
|
"type": "string"
|
|
3069
3069
|
},
|
|
3070
3070
|
"reasoningEffort": {
|
|
@@ -3834,7 +3834,7 @@ Classify a document or text into one of a fixed label set using an LLM. Output e
|
|
|
3834
3834
|
| `input` | string | yes | | Template expression for the text to classify. Typically the output of ai.parse, e.g. "{{ steps.parse.output.text }}". |
|
|
3835
3835
|
| `labels` | array<object> | yes | | Allowed labels. The LLM is constrained to pick exactly one of these names. |
|
|
3836
3836
|
| `prompt` | string | no | | Custom classification instructions appended to the system prompt. Use to clarify edge cases or emphasize evidence the model should weigh. |
|
|
3837
|
-
| `provider` | string | no | | Provider ID from eigenpal.config.yaml (e.g. "openai-
|
|
3837
|
+
| `provider` | string | no | | Provider ID from eigenpal.config.yaml (e.g. "openai/gpt-6-luna"). Falls back to the tenant default LLM provider when omitted. |
|
|
3838
3838
|
| `model` | string | no | | Model override (advanced) |
|
|
3839
3839
|
| `reasoningEffort` | `"none"` \| `"minimal"` \| `"low"` \| `"medium"` \| `"high"` \| `"xhigh"` \| `"max"` | no | | Reasoning effort for models that support it. Omit to use the selected model's default. |
|
|
3840
3840
|
| `maxInputTokens` | integer | no | | Max input tokens. Truncates input text when exceeded. Omit for no limit. |
|
|
@@ -3888,7 +3888,7 @@ Config schema:
|
|
|
3888
3888
|
"type": "string"
|
|
3889
3889
|
},
|
|
3890
3890
|
"provider": {
|
|
3891
|
-
"description": "Provider ID from eigenpal.config.yaml (e.g. \"openai-
|
|
3891
|
+
"description": "Provider ID from eigenpal.config.yaml (e.g. \"openai/gpt-6-luna\"). Falls back to the tenant default LLM provider when omitted.",
|
|
3892
3892
|
"type": "string"
|
|
3893
3893
|
},
|
|
3894
3894
|
"model": {
|
|
@@ -533,7 +533,7 @@ _Generated from `WorkflowDefinitionSchema` in `@eigenpal/types/src/workflow/work
|
|
|
533
533
|
| `values` | array<string> | no | | Closed set of allowed strings when type is enum. |
|
|
534
534
|
| `items` | object | no | | Element definition when type is array. |
|
|
535
535
|
| `properties` | array<unknown> | no | | Recursive field definitions when type is object. |
|
|
536
|
-
| `source` | string | no | |
|
|
536
|
+
| `source` | string | no | | Named external file source connection for single-tenant file-reference inputs, for example documents-s3; valid on type file or an array of file (one reference per file). |
|
|
537
537
|
| `mimeType` | string | no | | MIME hint such as application/pdf for a sourced file; mutually exclusive with extension. |
|
|
538
538
|
| `extension` | string | no | | Extension hint such as pdf for a sourced file; mutually exclusive with mimeType. |
|
|
539
539
|
|
|
@@ -794,7 +794,7 @@ This JSON Schema is generated from the same Zod schema used to parse workflow YA
|
|
|
794
794
|
}
|
|
795
795
|
},
|
|
796
796
|
"source": {
|
|
797
|
-
"description": "
|
|
797
|
+
"description": "Named external file source connection for single-tenant file-reference inputs, for example documents-s3; valid on type file or an array of file (one reference per file).",
|
|
798
798
|
"type": "string",
|
|
799
799
|
"minLength": 1
|
|
800
800
|
},
|