The file-first model is ending. Kin is the semantic system of record for AI-native software work.Learn more →
Built by Firelock

Firelock Labs.

The current Kin stack starts with the local semantic substrate, then moves up into hosted collaboration. The old editor and agent forks are retired from the active stack. The public story is Kin, KinDB, kin-vfs, Kin Editor, and KinLab.

Open source local substrateProprietary hosted layerSecondary systems research

Pre-registered determinism

26 / 26

Kin runs were bit-identical across all 26 tasks; the lexical baseline was bit-identical on 3 of 26.

Agent integration

60+ tools

Kin exposes semantic operations over MCP so any compatible agent can use the graph directly.

Public repo sync

GitHub-backed

Version badges use live GitHub metadata during deploys, with visibly labeled, date-verified fallbacks for offline builds.

Benchmark figures are scoped to one pre-registered 26-task Go suite on one model and runtime. They measure retrieval and localization, not patch success.Read the public proof summary.

Layer 1: Local Semantic Substrate

Lead with the local open-source stack: semantic truth in kin, graph storage in kin-db, transparent file projection inkin-vfs, and editor UX through kin-editor.

Coming Soon

Newly public supporting repos that are part of the wider Kin ecosystem story, but not yet presented here as full flagship product surfaces.

No Forks Required

The current Kin story is intentionally cleaner than the older material. Any MCP-compatible agent connects through Kin's built-in MCP server. Any editor can read projected files through kin-vfs. VS Code gets a focused extension for graph-native commands and navigation. The graph stays central; the surfaces stay replaceable.

Current access points

Editors: kin-vfs + Kin Editor
Agents: Kin MCP server via stdio
Storage: KinDB below the semantic graph
Teams: KinLab above the local substrate

Layer 3: Secondary Systems Track

Longer-horizon research comes last. It matters, but it should never displace the active local substrate plus hosted collaboration story.

Kin Kernel

Trusted semantic control plane

Secondary researchRust

Research into what process management, policy, and inter-actor coordination look like when semantic scopes are first-class instead of bolted on to file descriptors and PIDs.

Kernel researchPolicyExecution semantics

Kin OS

Semantic-first runtime model

Secondary researchRust

Long-horizon operating-system research for a world where semantic truth, not the filesystem, is the native substrate for software and automation.

Post-UnixRuntime designSemantic-native

Kin Hardware

Capability-aware hardware boundary

Secondary researchRust

Exploration of hardware that self-describes capabilities and participates in semantic coordination, rather than flattening everything into generic byte streams.

Hardware researchCapability modelSemantic topology

Browse the public repos

The public Kin repos live under firelock-ai. This page is aligned to the current public topology and GitHub tag state rather than the older fork-era product map.

Browse firelock-ai on GitHub