DataShield Corpus v0.6.3

Corpus: the agentic skills MCP server

You distribute agents. Do you know which customers run a fork? DataShield Corpus does — layered catalogs with drift detection for documentation, agent definitions, and skills: vendor versions, your customizations, their copies.

Corpus is an agent-facing MCP server, shipped on the production stack: every request metered and hash-chained through Auth.

Layered catalog with drift detection Blueprint design: a root catalog layer, a vendor customization layer, and per-customer copies with drift detection; usage meters back to Auth as a metering provider. In design — read the RFC. IN DESIGN — READ THE RFC L0 — Root catalog agent.yaml agent.yaml agent.yaml L1 — Vendor layer +patch your customizations · pinned versions L2 — Customer copies in sync in sync DRIFT: fork detected Δ tools[3], model pin removed token-denominated usage Auth federates as metering provider → same evidence plane (design) three layers, one diff

An MCP server for agent knowledge with a layered catalog

Corpus serves the artifacts agents run on — documentation, agent definitions, skills, prompts, tool specs — over MCP, with every request metered and hash-chained through Auth. It is not a dataset store; governed data lives in Ontology.

On top sits the layered catalog:

Three layers, one diff. When a customer forks — edits a skill, removes a model pin — that drift is detected and named, not discovered in an incident review.

What you get

Layered catalogs with drift detection

L0/L1/L2 catalogs keep vendor definitions, your customizations, and customer copies of docs, agents, and skills in one continuously diffed view. The design is public as an RFC — the implementation ships in Corpus.

Built on the production stack

Every Corpus request rides the shipped governance rails: Auth v1.1.8 per-tool-call metering and hash-chained audit. Governed data access stays in Ontology v0.24.1.

Early-adopter program

Early adopters get direct input on the metering-schema roadmap and a pricing lock at launch rates. Read the RFC first, then email support@myorg.ai.

Three questions before you adopt Corpus

Can I use Corpus today?

Yes. Corpus v0.6.3 is shipped: an MCP server for documentation, agent definitions, and skills, with the L0/L1/L2 layered catalog and usage metering that federates into Auth.

How does Corpus relate to Ontology?

Ontology governs data — classification, tokenization, analytics over tokenized datasets. Corpus governs agent knowledge — the documentation, agent definitions, and skills agents load — and tracks how deployed copies drift.

What does an early adopter actually get?

Input on the metering schema while the roadmap is still changeable, and a pricing lock at launch rates. In exchange we get your distribution and drift-detection requirements early.

Corpus v0.6.3 — shipped. Federates with Auth and Ontology, both in production.

The shipped stack — Auth, Ontology, Corpus, Guardian, Lighthouse — is priced and available now.

See pricing

You've seen the proof

Ready for a number? Scope your deployment and we'll price it against your own economics.

Get your quote →