Document Intake

Who the node is. The id is the key the DOT pins.

id

document-intake

name

Document Intake

type

tool

phases

none declared

What it does, and the prose the agent is handed when the graph runs.

action34 words

Open the run on a batch of unstructured source documents, stamp each one with its origin and content hash, and hand the batch downstream as the unit of work everything else is keyed to.

spec115 words

Collect the source documents waiting in the intake location and open a run over them, in batches of at most 100 documents. For each document record three things, the origin path it came from, its full content, and a stable hash of that content, and emit the batch on documents as one entry per document. Use that hash as the document's identity for the rest of the run: bytes you have seen before must hash to the same value, because that is what lets a re-run land on the record already written instead of a second copy of it. You are done when every document in the intake location appears in exactly one emitted batch.in full above

model

whatever the graph supplies

agent

not named

skill

skills/document-intake.md

tools

none

mcp

filesystem

params

batch_size: 100

What arrives, what leaves, which nodes it expects to hear from, and what may not.

inputs

none

outputs

documents : json

dependencies

none

cannot

no type is refused

will_not

emit the same document in two batches

The keys the static analysis reads. Nothing here instructs the agent.

risk_markers

none

notes30 words

The content hash is what makes a re-run idempotent: a document that arrives twice resolves to the same record key at the store and upserts over itself rather than duplicating.

The card's own version, and who wrote it.

version

1.0.0

author

autogen

provenance

not stated