Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.appliedaifoundation.org/llms.txt

Use this file to discover all available pages before exploring further.

​
What it does

The Vessel Intelligence skill is the read/write interface to the structured knowledge base behind every vessel: cases, evidence, observations, agent notes, and learnings. It lets agents and humans query “what do we know about X?” and log new findings without touching raw files. The intelligence layer keeps three distinct types of records, each with its own lifecycle:
LayerWhat it capturesWho writes it
ObservationsFactual statements about a case (“seawater leak damaged motor windings”)Any agent
NotesOpinions, recommendations, flags (“similar risk on sister ships”)Any agent or human
LearningsWhat the fleet should change (“install leak detection near motors”)Senior agents
Every record links to evidence — case folders, files, emails — so reasoning can always be traced back to source.

​
When to use it

  • “What do we know about this case?”
  • “Log this finding”
  • “Search vessel observations”
  • “Write a learning”
  • “Get all intelligence for AQUILA”

​
What you can query

  • Every case for a vessel with status, priority, and recent activity
  • All observations for a case
  • All notes targeted at a vessel, case, or piece of equipment
  • All learnings derived from a case, vessel, or equipment family
  • Cross-vessel patterns when a learning has been promoted fleet-wide

​
What you can write

  • A new observation linked to a case (factual, with evidence)
  • A note on a case, vessel, or equipment (opinion, with reasoning)
  • A learning derived from a case (recommendation, optionally promoted fleet-wide)

​
How it works

The skill operates against a per-vessel intelligence database plus a fleet-wide database. Reads and writes go through a query runner — there is no need to write SQL by hand. Every write captures the author, timestamp, evidence references, and the case lineage.

​
Why three layers

Mixing facts, opinions, and recommendations in one bucket destroys their usefulness. Splitting them lets the system answer different questions cleanly:
  • “What happened?” — observations
  • “What does the team think about this?” — notes
  • “What should we do differently?” — learnings
Promoting a learning fleet-wide is the moment a single-vessel finding becomes a fleet improvement.
  • search-vessel-cases (now removed from docs) — historical case search
  • search-learnings (now removed from docs) — fleet learning search
  • status-rollup — fleet-wide aggregation that consumes intelligence records