Photo ingest execution loop

Telegram and OpenClaw image batches should flow into Shanepedia and the relevant child project lanes as durable operator inputs. The goal is not just to archive photos. The goal is to turn them into better context, better ideas, and better actions.

Canonical flow

1. Find the Telegram session memory entry and attached media paths.
2. Create a raw source manifest in raw/sources/ with message IDs, context, and file paths.
3. Sample or analyze the images and classify each one:

• place
• contact
• operator UI
• product
• inspiration
• business lead

1. Write a wiki source-summary page describing the coherent signals in the batch.
2. Update relevant founder pages, project pages, and contacts.
3. Generate action items:

• add contact
• draft business idea
• create follow-up page
• prepare outreach copy
• slot into future-trip planning

What the current photo batches prove

The 2026-04-04 batches already show the system should handle:

• product packaging and brand inspiration
• finca/agro context
• operator mobile sales pages
• WhatsApp conversation screenshots and lead capture

That is enough to support trip-planning, local-operator work, and broader business-idea generation.

Execution rule

Every photo batch should end in at least one durable artifact and one actionable next step. If the images only get summarized in chat, the system is failing.