Delta Intake

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

id

delta-intake

name

Delta Intake

type

tool

phases

none declared

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

action30 words

Open the nightly run on the window since the last successful watermark and hand that window downstream, so a missed night widens the next run rather than losing a day.

spec116 words

Read the last_successful_run watermark and work out the window this run has to sweep: it starts at the watermark and ends at the moment you run. Emit that window on window as an explicit start and end timestamp in UTC, together with the watermark value you derived them from. Never clip the window back to the nominal 24 hours when the watermark is older than that, a night that did not run has to widen the next window, not quietly lose its rows. If there is no watermark at all, emit a window covering the last 24 hours and mark it as a cold start so the reader knows the bound was assumed rather than recorded.in full above

model

whatever the graph supplies

agent

not named

skill

skills/delta-intake.md

tools

none

mcp

none

params

schedule: 0 2 * * *, watermark: last_successful_run, window_hours: 24

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

inputs

none

outputs

window : json

dependencies

none

cannot

no type is refused

will_not

clip the window when the watermark is older than it

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

risk_markers

none

notes29 words

The schedule lives here as a parameter because the core vocabulary has no scheduler capability to declare; the trigger is a cron entry in every sense except the ontology's.

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

version

1.0.0

author

autogen

provenance

not stated