WormVolt

Inside WormVolt

How a living
knowledge vault works.

Editable knowledge. Configurable autonomy. Background information flows. This is the architecture we’re developing to connect the work you do with the knowledge you keep.

Read the current development status →
Architecture illustration · Simulated behaviour

Follow the knowledge.

Useful work becomes a richer brain.

A research result joins working knowledge, refines an existing note and connects to a project. Source material and AI suggestions remain distinct.

Source

Research finding

Source material from a connected task

Your maintained brain
project.md
decisions.md
research.md
Destination

Project knowledge

Notes, decisions and related research

  1. Source added
  2. Existing note refined
  3. Related knowledge connected

A project note, ready to develop.

Your task continues. Knowledge moves.

A captured change can start permitted work in the background: refine knowledge, prepare selected context and update a connected destination.

Independent activity

Current task

Continue your conversation or work

Current task continues
Your maintained brain
project.md
decisions.md
research.md
Destination

Connected vault

Receives selected, transformed context

  1. Change captured
  2. Working knowledge updated
  3. Selected context prepared
  4. Destination updated

Ready to capture a change. Your current task continues independently.

Keep the brain. Adapt the connection.

Replace a tool without discarding your accumulated notes, organisation or definitions. A suitable adapter prepares the context for its new destination.

Source

Your knowledge

The notes and structure stay

Your maintained brain
project.md
decisions.md
research.md
Destination

Tool A

Adapter A · destination format

  1. Knowledge retained
  2. Connection adapted
  3. Tool B receives selected context

Tool A connected in this illustration.

01 / Knowledge lifecycle

Knowledge develops
through use.

A living vault holds more than facts: preferences, reusable instructions, procedures, project history and working context can all belong in the knowledge foundation.

Useful information can refine an existing record or reveal a better way to organise it. Selected context supports another task; useful outcomes can return.

  1. Capture useful information from connected work.
  2. Add, revise or reorganise relevant knowledge.
  3. Retain earlier material as versions or archives where defined.
  4. Prepare selected context for another task.
  5. Feed useful outcomes back into the knowledge.

Source material and AI suggestions need appropriate treatment. More text does not automatically mean better knowledge.

02 / Background flows

Work continues.
Knowledge moves.

A change can initiate background work under the source volt’s definition. A connection selects and transforms permitted information, and a receiving volt can perform its own allowed activity.

This can include refinement, archiving, new versions or further research. The user continues the current task; the system asks for direction where the definition or an exception requires it.

Two triggers. One foundation.

During ongoing work: a useful change starts background capture and propagation.

When an agent starts: the task requests the context it needs from maintained knowledge.

Engineering questions being established

Conflict resolution, retries, loop prevention, stale downstream context and correction need explicit behaviour. An agent that depends on an update needs a readiness check. Asynchronous operation is not a guarantee of instant consistency.

03 / Definitions

Autonomy belongs
to each volt.

A volt is a knowledge collection with a definition governing its purpose, agents, permitted activity and connections. “Self-governing” means operating within that definition.

Some definitions keep a source unchanged and allow only outgoing updates. Others permit frequent refinement. Review can apply to particular actions without interrupting every update.

01

Stable reference

Keep the source unchanged while allowing selected outputs.

02

Evolving working knowledge

Allow defined additions, revisions and organisation in the background.

03

Reviewed sharing

Request review before sending selected changes onward.

Example configurations, not fixed product types. A note link does not grant access. Enforcement, visibility and recovery must be established in the implementation.

04 / Connections & transformation

Prepare knowledge
for its next use.

Portability goes beyond exporting a folder. Another tool might need a selected brief, a different structure or a derived view. A connection prepares appropriate information for that destination.

The flow can work in both directions: a specialist tool may produce useful knowledge or a better representation to bring back into the vault.

An intended connection
  1. Select relevant, permitted knowledge.
  2. Restructure or transform it for the destination.
  3. Carry permitted updates onward.
  4. Bring useful results back where allowed.

Use suitable existing connectors or develop new ones. The connector inventory and extension interface are still being defined.

05 / Context for agents

A useful starting point
for the next task.

A new agent should be able to start from maintained knowledge instead of a manually rebuilt brief. Context preparation during provisioning is part of WormVolt’s direction, alongside continuous background activity.

Clear responsibilities

WormVolt: maintain, transform and prepare the knowledge.

KrakenBot: the intended integration for provisioning and deploying agents.

The integration needs defined readiness and failure behaviour. The wider architecture aims to support appropriate connections to other tools as well.

Build with us

Help shape a more useful
foundation for AI.

Explore the architecture and help turn portable, evolving knowledge into an everyday tool.

Explore the project Read the technology overview