Hyperconsciousness

Agent-enablingOpen Source

Hyperconsciousness gives agent builders a self-hosted knowledge store for carrying scoped context and task handoffs across sessions. Agents can retrieve authorized records, capture notes with source references, and correct or retract managed notes while append-only history remains intact. It suits developers experimenting with inspectable memory who can operate an alpha-stage system and its security boundaries.

Encrypted, append-only knowledge storage with scoped MCP access and persistent task handoffs for agents.

Visit website GitHub

Overview

About Hyperconsciousness

Hyperconsciousness (hc) is a developer-alpha knowledge engine for humans and agents. Its MCP interface retrieves authorized context and accepts captured notes when the caller has write access. Managed notes support explicit versions, corrections, and retractions while preserving append-only history.

So agents can…

Search stored context and reopen selected records under scoped, expiring grants.
Persist a task handoff with decisions, unresolved questions, and declared source references for a later session to reopen under its own access.
Apply explicit versioned corrections to managed notes without treating superseded text as current context.

Start with the repository’s source installation instructions and an isolated demo store. HC does not assemble the model’s context window or execute the agent’s tasks. Encryption at rest and during replication does not hide returned plaintext from the receiving model provider.

Listing decision

Why Hyperconsciousness is listed

Its memory workflow includes structured retrieval, durable record references, and versioned corrections or retractions of managed notes. These capabilities support continuity across harness sessions rather than only exposing a generic database through MCP.

Before you choose

Fit and limitations

Best for

Developers experimenting with persistent, inspectable agent context in a locally operated store.

Not best for

Users seeking a hosted agent runtime or a production-mature, independently audited memory service.

Known limitations

Developer alpha: commands and APIs may change. Grants constrain server responses, not processes with access to the owner account or keys. Hosted model providers can see returned plaintext. Retractions do not erase historical ciphertext or copies already made. No independent security audit is claimed.

Not yet verified

No independent performance or security evaluation was performed for this submission.

Sources

Sources and verification

First-party unless marked otherwise. Each link supports only its adjacent claim.

Repository README
Rust engine, CLI, MCP/HTTP interfaces, encrypted append-only records, device sync, scoped expiring grants, source build instructions and developer-alpha status.
official repository · accessed Sep 28, 2026
Capture context and write receipts
Structured retrieval, reopenable record references, task handoffs with declared source references, and versioned corrections/retractions for managed notes. Handoffs do not transfer access.
official documentation · accessed Sep 28, 2026
Threat model and constraints
Trusted-node and process-access boundaries; revocation cannot recall previously copied plaintext.
official documentation · accessed Sep 28, 2026
MIT license
The public software is MIT licensed and copyright Louis Beaumont.
official license · accessed Sep 28, 2026
Mission and repository boundary
Knowledge storage and access are separate from agent orchestration, model routing and tool execution.
official documentation · accessed Sep 28, 2026