---
name: process-inbox
description: Process accumulated inboxes, lists, or other series in a dedicated session using their local lifecycle. Use for classifying and triaging raw material into maintained notes, not for capturing one incoming message or auditing without edits.
---

# Process a knowledge inbox

Read [knowledge-rule](../knowledge-rule/SKILL.md), then the storage's lifecycle
and relevant folder readmes. Follow its distinction between dated incoming
artifacts and disposable raw inbox. If that distinction is missing, propose
it before doing destructive processing.

Before changing sources, inventory the items with their source locations or
existing identifiers. Build and review a temporary view grouped by similarity,
including an unresolved section. Classification changes nothing.

Process one section at a time. For each item, find an existing canonical object
before creating another. Update an existing paragraph or extract an independent
object as appropriate; carry its provenance and relevant context. Use
[move](../move/SKILL.md) when processing relocates knowledge and retires its
inbox source; it adds [cross-boundary](../cross-boundary/SKILL.md) when material
belongs elsewhere.

A section is not a unit of deletion: retire an item only through `move` after
verifying every destination, including when one item produces several objects.
Keep durable source artifacts. Retain the temporary view until it reconciles
verified destinations with unresolved or failed sources, then report
source-to-destination results by section.