Get card
Download
field-extractor@1.0.0.yamlClone
npx -y darkprint clone field-extractor@1.0.0There is no repository and no history behind a card. Download hands you the document as it stands, and Clone fetches the same document by name.
Read each source document and pull the target fields out of it verbatim, quoting the span each value came from so a downstream reviewer can check the extraction rather than trust it.
used in 1 blueprint
Specification
138 words · handed to the agentFor every entry on documents, read the document's content and pull out the target fields named in the extraction schema you are configured with. Copy each value verbatim from the source text, do not paraphrase it, do not tidy it, do not convert a unit or a date format, and record next to it the span of the document the value was taken from, so a reviewer can check the extraction instead of trusting it. Emit one object per document on fields, keyed by field name and carrying that document's hash, so a later step can key back to the source. Where the document simply does not state a field, leave it absent: an absent field is a fact the next step can act on, and an invented one is a plausible record rather than a true one.
Interfaces
1 in · 1 outInputs
1| Name | Data type | Required | Description |
|---|---|---|---|
| documents | json | required | The intake batch, one entry per source document, with its content and hash. |
Outputs
1| Name | Data type | Description |
|---|---|---|
| fields | json | One object per document, field name to extracted value, with the source span. |
Dependencies
1The upstream nodes this card expects to receive from. Whenever a blueprint pins this card, each name is checked against a real edge in that graph. A name without a link is not a published card; it refers to a node inside some graph.
Card values
17 declaredWho the node is. The id is the key the DOT pins.
- id
- field-extractor
- name
- Field Extractor
- type
- agent
- phases
- Implementation
What it does, and the prose the agent is handed when the graph runs.
- action32 words
- Read each source document and pull the target fields out of it verbatim, quoting the span each value came from so a downstream reviewer can check the extraction rather than trust it.
- spec138 words
- For every entry on
documents, read the document's content and pull out the target fields named in the extraction schema you are configured with. Copy each value verbatim from the source text, do not paraphrase it, do not tidy it, do not convert a unit or a date format, and record next to it the span of the document the value was taken from, so a reviewer can check the extraction instead of trusting it. Emit one object per document onfields, keyed by field name and carrying that document's hash, so a later step can key back to the source. Where the document simply does not state a field, leave it absent: an absent field is a fact the next step can act on, and an invented one is a plausible record rather than a true one.in full above - model
- claude-haiku-4-5
- agent
- Extractor
- skill
- skills/field-extractor.md
- tools
- none
- mcp
- none
- params
- max_tokens: 2048, temperature: 0
What arrives, what leaves, which nodes it expects to hear from, and what may not.
- inputs
- documents : json
- outputs
- fields : json
- dependencies
- document-intake
- cannot
- no type is refused
- will_not
- paraphrase a value it copies, invent a field the document does not state
The keys the static analysis reads. Nothing here instructs the agent.
- risk_markers
- none
- notes30 words
temperature: 0is not a style preference here. Extraction is a transcription task, and a model that paraphrases a value is producing a plausible record rather than a true one.
The card's own version, and who wrote it.
- version
- 1.0.0
- author
- autogen
- provenance
- not stated
Version history
1 version published- field-extractor@1.0.0currentsha256:4fe31039c125ad69dede4ae26dd86b766dbda7179a13d76ac4ca569045b68927
pinned bySchema Forge ETL
autogen/schema-forge-etl
A digest is a fingerprint (SHA-256) of the card's content, computed without the author and provenance fields. The same card from two people gets the same digest; any edit gets a new one.
First published version, so there is nothing to compare yet. Versions are never edited in place: the next change arrives as a new version, and the differences between the two documents are listed here.
Community notes (0)
No notes yet.
Nobody has posted about this node card yet.
Sign in to post a note.