Local-first context for AI

Stop re-explaining yourself across AI tools.

Holoself keeps approved personal context in readable local files, links it to independent projects through task-specific lenses, and asks before project discoveries become durable knowledge.

Open source· Markdown-first· No hosted account· No silent memory updates
Available now · v0.8.0
You
Prepare this interview using my career evidence, but keep compensation and publishing-private context out.
AI + Holoself
Career lens loaded. I found approved claims from their source files, excluded restricted material, and preserved provenance. New learning will be a proposal—not a silent update.
New in 0.8.0

Context where the work happens.

Supported AI clients can request bounded context inside an already linked project, on demand—without a daemon, hosted account, or second source of truth.

One reviewed setup

Preview and configure project-bound MCP for Codex, Claude Code, and supported AGY projects from one local workflow.

Load less, with evidence

Typed status, manifest, context, and search tools return structured, size-bounded results with provenance.

Control stays explicit

The project link remains the authority. MCP cannot broaden a lens, approve proposals, or write canonical context.

Fallbacks stay first-class

The Workbench, CLI, shared skill, BOOTSTRAP, and reviewed snapshots remain deterministic ways to work.

Whole-person context protocol

One whole person. Relevant context for each project.

Lenses change relevance, depth, and privacy—not identity.

Self

Approved reusable context

Identity, preferences, voice, evidence, decisions, stories, and domain context.

Lens

Task-specific selection

Career, publishing, technical, leadership, interview, general, or private.

Project

Independent execution

Applications, posts, research, code, and other project-owned artifacts stay local.

Project discovery → evidence-backed proposal → your review → approved reusable knowledge
Ownership by design

Projects keep their work. You keep reusable knowledge.

Self owns

Identity, preferences, durable facts, approved stories, decisions, disclosure rules, and evidence.

Projects own

Applications, research, interview notes, posts, calendars, performance data, templates, and execution strategy.

Why it matters

Continuity without a central vault.

Less repetition

Reuse approved context

Bring your voice, preferences, evidence, and decisions into new work without copying a complete profile into every project.

Clear boundaries

Keep projects independent

Career and publishing workspaces can read the same self context while retaining their own artifacts and private details.

Governed learning

Review before memory

Holoself shows target, evidence, affected files, and proposed diff before an approved canonical update.

Use cases

Same person. Different lens. Clear provenance.

Career execution

Use approved achievements and leadership stories while keeping job research, compensation, and interviewer strategy project-owned.

Explore career use case

Publishing

Retrieve public-safe claims and approved stories while filtering employer-confidential or career-private material.

Explore publishing use case

Technical work

Carry decision principles and technical preferences into architecture or coding projects without centralizing their source artifacts.

Explore technical use case
Trust guarantees

Readable source. Disposable indexes. Explicit change.

Local-first

No hosted account or required external service.

Markdown canonical

Generated packets and indexes never become source of truth.

Privacy-aware

Visibility, sensitivity, public-safety, and field restrictions filter output.

Provenance

Resolved context and accepted claims identify their source files.

Tool adapters

Use context where work happens.

Project-bound MCP gives verified clients typed, on-demand access to the same lens, privacy, lifecycle, and provenance rules as the CLI. Codex and Claude Code discovery were verified for 0.8.0; AGY verification remains bounded by its project registration and approval policy.

Current maturity: open-source preview. The holoself-ai@0.8.0 CLI package is published on npm; a reviewed source checkout remains available for development and audit.