Use “Match a dropped document” in Recued. Fires when a file lands in the relevant record and proposes which outstanding checklist line it satisfies. It proposes only — the line stays missing until you confirm it.
Match a dropped document
Use “Match a dropped document” in Recued. Fires when a file lands in the relevant record and proposes which outstanding checklist line it satisfies. It proposes only — the line stays missing until you confirm it.
Complete workflow
Client Document Intake
9 pinned recipes
Use Client Document Intake in Recued for drop link, reception, documents, and checklist. It includes 14 built-in actions and 8 ready-to-run workflows. Actions that change data use Recued's approval controls.
You are viewing one recipe in this workflow. Installing it opens the complete pack so its other recipes and background automations arrive together.
Use “Open a document request” in Recued. Open one document request against a Reception drop link and write its checklist.
Use “Confirm a document” in Recued. Accept the proposal on one checklist line, or clear it. This is the step that turns a file arriving into a document received — the two are different facts and the pack keeps them apart.
Use “Accept documents in bulk” in Recued. Tick the proposals you believe and accept them in one go. Same decision as confirming one line, made once for several — the board stays the board, and this is where a batch is done.
Use “Repair blank checklist lines” in Recued. Remove the empty checklist lines a defect wrote before 21 September 2026, so a request opened with no documents stops showing a line whose label is a leftover placeholder. It asks before each removal, and it never deletes the request itself.
Use “Document requests board” in Recued. Every open document request with what is still outstanding, and every line waiting on your decision. It reads information without changing data in the connected service.
Show all 9 recipes
Use “Chase the missing documents” in Recued. Email one client the list of documents still outstanding, with their drop link.
Use “Records runtime” in Recued. It helps you complete this workflow using the connections and settings you choose.
Use “One required doc” in Recued. Open one required doc from the list that names it, and read everything recorded about that required doc on one page.
What it does, and what it deliberately does not
On every file arriving in data.file.received it reads the filename and MIME type, looks at the checklist lines still outstanding across your OPEN requests, and asks the model which single line the upload best satisfies. The winning line records a proposal: the file id, the filename, and a confidence. Its status becomes proposed.
Nothing is marked received. Confirming is Confirm a document — one click from the board.
⛔ It cannot tell where a file came from. core.storage.data-file-read returns record_id, filename, mime_type, size_bytes and blob_hash — no origin, no source endpoint — and a file event carries no record projection to re-read it from. So a mail attachment arriving while a request is open is indistinguishable from a client's upload. Proposing rather than deciding is what makes that safe.
⛔ Only the filename and MIME type are read. The file's CONTENT is never opened. Reading bytes is a gated egress boundary, and a document you have not accepted yet is exactly the wrong thing to feed a model. If the filename is generic (scan.pdf, IMG_1234.jpg) expect no proposal — that is the intended outcome, not a failure.
⚠ Uploads are unscanned by default. A drop ingests as scan_status: 'unscanned' and the substrate's gate is advisory. Install a scanner pack (clamav-scan / defender-scan) if you want that verdict to mean something before you open anything.
⚠ A file can only be proposed against one line at a time, and re-proposing a line that already carries a proposal is skipped — the first proposal stands until you accept or clear it.
How it works
Inspect the data fetches, transforms, gates, and output this recipe runs.
Process (23 steps)
record_id
default
Apply default
has_record
compare
Check if record id has data
record_guard
guard
Stop if has record equals
outstanding
recued-core.client-document-intake.required_doc.search
Search required doc on client document intake
outstanding_rows
default
Apply default
outstanding_count
count
Count items in outstanding rows
has_outstanding
compare
Check if outstanding count is greater than 0
nothing_outstanding_guard
guard
Stop if has outstanding equals
file_meta
core.storage.data-file-read
Data file read storage
defaults
defaults
Apply defaults
signals
merge
Combine objects together
candidate_labels
pluck
Extract label from each item
classify
core.ai.classify
AI classifies data into one of the given categories
defaults_2
defaults
Apply defaults
confident_enough
compare
Check if confidence is at least setting min confidence
matched_rows
filter
Filter by condition
matched_count
count
Count items in matched rows
matched_exactly_one
compare
Check if matched count equals 1
defaults_3
defaults
Apply defaults
already_proposed
compare
Check if existing proposal has data
should_propose
all
Check if all conditions are true
propose
recued-core.client-document-intake.required_doc.update
Update required doc on client document intake
skip: step.should_propose equal false
card
to_summary
Format results as a summary card
▼
Output
summary
card
Settings
Configurable at install. Defaults shown — change them anytime in Recued.
min confidence
setting
=
0.6
Uses (4 operations)
Pack operation
Search required doc on client document intake
recued-core.client-document-intake.required_doc.search
Files
Data file read storage
core.storage.data-file-read
AI models
AI classifies data into one of the given categories
core.ai.classify
Pack operation
Update required doc on client document intake
recued-core.client-document-intake.required_doc.update
Trust & control
What installing this recipe would let it do. Recued grants these permissions at install — review them there before approving.
What it's allowed to do
Write
Pack
Client Document Intake
external change
Read: required_doc.search · No approvalWrite: required_doc.update · No approval
Read
Files
core.storage.data-file-read
read-only
Read
AI models
core.ai.classify
read-only