Technology / Architecture in development

How a living knowledge vault works.

WormVolt brings together editable knowledge, configurable autonomy and background information flows. This overview explains the architecture being developed.

Project resources ↗

Architecture illustration · Simulated behaviour

One collection. Three possibilities.

Explore how knowledge grows, moves and stays useful as tools change.

Your knowledge

Markdown · Obsidian

▤ project.md

Export behaviour: to investigate.

Working note · original
▤ decision.md

Keep knowledge portable.

Related research can inform this decision.
▤ research.md

Portability findings

Definition retained

Project research · scoped automatic changes · selected outgoing summaries

Grow knowledge

Source material

Research result: the export preserves Markdown links.

AI suggestion

AI suggestion: connect the portability decision to this research.

Illustrated outcome: revise the project note using the source, then connect the decision and research. The suggestion stays identified as a suggestion.

Useful work can develop the knowledge and its structure.

Sync in the background

Current taskContinue drafting the project plan

Your task has its own lane.

Background activity
  1. Captured
  2. Updating working knowledge
  3. Preparing selected context
  4. Destination updated

Permitted background activity can continue while the current task moves on.

Change a tool

Adapter and destination format

Adapter A → structured task brief

Relevant context ↓Tool A

Swap to Tool B with a concise Markdown context adapter. The same notes, organisation and definition stay in the central collection.

The accumulated knowledge stays. The connection is adapted.

Choose a view to explore. No tools are connected.

01 / Knowledge lifecycle

Knowledge develops through use

A volt is a knowledge collection with a definition for how it behaves. Facts, preferences, reusable instructions, procedures and project history can all contribute to its working context.

Useful information enters through connected workflows. Relevant knowledge is added, revised or reorganised; earlier material may be retained as versions or archived. Selected context is prepared for another task, whose useful outcomes may contribute back.

Source material and AI suggestions need appropriate treatment. A suggestion should not silently become an established fact. More text is not automatically better knowledge: relevance, provenance and corrections matter. Capture depends on the connection and the chosen definition.

Architecture illustration
knowledge / projects
▤ project.mdKeep research close to decisions.

A maintained project note links what we know to why we chose it.

↳ decision.md Why Markdown?
↳ research.md Portability findings
BeforeResearch filed separatelyAfterProject → decision ↔ research

An illustrative structure: revise the notes and connect the reasoning.

02 / Background flows

Work continues while knowledge moves

A change can initiate background activity while the current task continues. The volt’s definition determines the permitted work: revising knowledge, retaining a version, archiving earlier material or preparing an output.

A connection may transform information for its destination. That update can initiate further permitted activity in another volt. Some actions may ask for direction or approval; others can proceed automatically within their defined scope.

Asynchronous work continues independently. If an agent launch depends on a particular update, it needs defined readiness behaviour: wait for the required context, use an explicitly acceptable earlier version, or ask for direction. These are design choices, not a promise of instant consistency.

Engineering questions

Conflict handling, retries, loop prevention and visibility are still being established. Implementations need to define how concurrent revisions are resolved, how failed work is retried, how feedback cycles are bounded, and how people inspect progress and recover from errors. Required-context readiness must be explicit before relying on an agent launch.

03 / Definitions

Autonomy is configured per volt

A definition establishes the purpose of a volt, the allowed agents and operations, whether its source may change, what can move to each destination, and when approval or user input is required.

Stable reference knowledge can supply context without changing its source. Evolving working knowledge can accept scoped background revisions. Reviewed sharing can require a decision before publishing selected outputs. These are example configurations, not mandatory product classes.

Updating knowledge, using it as context and sharing it onward are separate decisions. A note link does not itself grant permission. Definitions express the desired boundaries; real implementations must enforce them appropriately.

Architecture illustration
A working knowledge definition
Purpose
Project research
Knowledge changes
Automatic within defined scope
Outgoing context
Selected summaries
Public sharing
Review required

Illustrative definition—not configuration syntax

04 / Connectors

Transform knowledge for its next use

Portability goes beyond exporting files. A connector selects relevant information and restructures or summarises it for a destination. It can carry permitted updates onward and bring useful representations or results back.

A task brief for an agent and a readable research summary may draw from the same collection, in different forms. Retaining the notes, their organisation and their definitions helps new tools build on existing work.

Developers should be able to reuse suitable connectors or develop new ones, extending the knowledge layer as tools evolve. The connector inventory and extension interface are still being defined.

Select → Transform → Supply → Feed back

05 / Agent context

Prepare context when an agent needs it

A new task should be able to start from maintained knowledge rather than a manually rebuilt brief. WormVolt’s direction includes context preparation when agents are provisioned, alongside ongoing background flows.

The intended flow selects knowledge relevant to the task, prepares the permitted representation and makes its readiness explicit. Useful outcomes can then return to the knowledge lifecycle.

KrakenBot handles the intended provisioning/deployment integration. WormVolt maintains and prepares the knowledge. This division leaves room for other future connections; KrakenBot is not a requirement for every possible adapter.

Maintained knowledge → Selected context → Agent task

Built to grow beyond one tool

Help shape a more useful foundation for AI.

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