·LIVE
Pro
·INDIVIDUAL ENGINEER
$15/mo
FREE TIER · NO CARD REQUIRED
- Managed Kvendra KB (Enterprise engine, hosted)
- Vault backup + profile sync across machines
- Cloud embeddings included
- Skills: recommended setup
- CLI broker: optional
§ 00 — POST-COPILOT ENGINEERING
Structure.
Not memory.
The harness for advanced engineering.
Kvendra is a system for professional software engineering with AI — not a copilot, not a chat cache. Typed knowledge base the AI can reason about. CLI vault that keeps secrets out of the LLM. Skills that orchestrate it all from Claude Code. Open-core, self-hostable, hosted Pro available.
SELF-HOST · FREE · LIVE
Run the whole stack on your hardware.
Platform · CLI vault · Skills — open source. Docker compose, your data, your governance.
GITHUB / KVENDRAAI HOSTED PRO · LIVE
≤3 min from signup to first MCP call.
Managed KB, vault backup, profile sync. Same wire, same MCP — $15/mo.
START PRO · BUILT FOR INDIVIDUAL ENGINEERS · TEAM WORKSPACES SOON
·V0.6 · 2026 ·INSTRUMENT-GRADE · SHIPPING
AI alone isn't enough.
Harnesses aren't either.
·01 / CONTEXT
Context vanishes.
Every new chat starts blank. Your AI rediscovers the same architecture, the same conventions, the same decisions — every session. Complex systems need memory the model can read structurally, not paste into a prompt.
·02 / GOVERNANCE
Harnesses bolt on. They don't structure.
Prompt-stacking, tool wrappers, agent frameworks — useful for toys. Engineering at scale needs typed contracts, versioned decisions, releases linked to requirements. Not chat history flattened into vectors.
·03 / SECURITY
Secrets leak through the LLM.
Every API key, every credential piped into model context is one screenshot away from the wrong place. Privileged actions belong outside the model. The LLM should never see what it doesn't need to know.
One system.
Three pillars.
·HOW IT WORKS
·LLM
Opus · Ollama
your IA model of choice
MCP
·KNOWLEDGE BASE
Typed entities
components · contracts · decisions · releases · runbooks
·ORCHESTRATOR - Claude Code
Skills
18 typed pipelines · open source
·CLI VAULT / BROKER
Zero-knowledge secrets
privileged actions · signed audit chain · LLM never sees keys
- ·01 / KNOWLEDGE BASE
A graph, not a vector blob.
·KVENDRA PLATFORM · AGPL-3.0Components, contracts, decisions, releases, runbooks, requirements — schema'd entities with relations and a versioned changelog. Mutations land through a transactional pipeline (begin · stage · commit) so every change is grouped, traceable and reversible. Hybrid search: semantic + typed filter.
·14 MCP TOOLS ·TRANSACTIONAL ·HMAC AUDIT ·PGVECTOR - ·02 / CLI VAULT
Secrets outside the model.
·KVENDRA CLI · APACHE-2.0A Rust binary that owns a zero-knowledge vault. The LLM never reads keys. Privileged commands (cloud, deploy, db) run through the broker — credentials stay on disk, every action is signed into a chained audit log.
·ZERO-KNOWLEDGE ·SIGNED LOG ·CROSS-PLATFORM - ·03 / SKILLS
The orchestrator.
·KVENDRA SKILLS · APACHE-2.018 typed pipelines for Claude Code. Each /slash command opens a transaction, reads the KB, runs privileged work through the vault, and commits results back as typed entities. Every change is atomic and audited — the AI never freelances.
·18 PIPELINES ·CLAUDE CODE ·MCP-NATIVE
Open from day one.
Free forever, self-hostable.
·PACKAGE / KB ENGINE
Kvendra Platform
AGPL-3.0
·PACKAGE / CLI VAULT
Kvendra CLI
APACHE-2.0
·PACKAGE / SKILLS
Kvendra Skills
APACHE-2.0
·QUICKSTART · SELF-HOST
cli · platform · skills
·01 CLI vault ·KVENDRA CLI · APACHE-2.0
The Rust binary that owns your zero-knowledge vault and brokers privileged commands. Install once; the LLM never sees your secrets.
$
cargo install kvendra ·MORE INSTALL OPTIONS docker · embeddings · MCP →
·02 Platform (KB engine) ·KVENDRA PLATFORM · AGPL-3.0
The typed KB needs an embeddings provider. Pick a route — the docker image accepts either a hosted API key (free tier) or a local Ollama endpoint.
- ·01 Create a free Kvendra account. SIGN UP
- ·02 Generate an embeddings API key from the dashboard. (200k tokens / month included, no card required)
- ·03 Inject it into the platform container and run.
$
export KVENDRA_EMBEDDINGS_API_KEY=sk_kv_…$
docker pull kvendra/kvendra-platform:latest$
docker run -d -p 8080:8080 --name kvendra \
-e KVENDRA_EMBEDDINGS_PROVIDER=hosted \
-e KVENDRA_EMBEDDINGS_API_KEY \
kvendra/kvendra-platformRun a local embedding model with Ollama and point the platform at it. Heavier on local hardware — everything stays on your machine, no signup required.
$
ollama pull nomic-embed-text$
docker pull kvendra/kvendra-platform:latest$
docker run -d -p 8080:8080 --name kvendra \
-e KVENDRA_EMBEDDINGS_PROVIDER=ollama \
-e KVENDRA_OLLAMA_BASE_URL=http://host.docker.internal:11434 \
-e KVENDRA_OLLAMA_MODEL=nomic-embed-text \
kvendra/kvendra-platformSpin up the platform with a deterministic mock embedder — zero network, zero signup. Use it for CI pipelines, local development, demos or schema work. Search ranking is not meaningful, structure is.
$
docker pull kvendra/kvendra-platform:latest$
docker run -d -p 8080:8080 --name kvendra \
-e KVENDRA_EMBEDDINGS_PROVIDER=mock \
kvendra/kvendra-platform ·FULL ENV REFERENCE · POSTGRES · COMPOSE hub.docker.com/r/kvendra/kvendra-platform →
·03 Skills (Claude Code plugin) ·KVENDRA SKILLS · APACHE-2.0
Add the marketplace once, then install the 18-skill plugin. Run inside a Claude Code session.
$
/plugin marketplace add KvendraAI/kvendra-skills$
/plugin install kvendra-skills@kvendra-skills ·MARKETPLACE github.com/KvendraAI/kvendra-skills →
·SOURCE / GITHUB
github.com/KvendraAI
·IMAGES / DOCKERHUB
hub.docker.com/u/kvendra
·READ / DOCS
/docs · architecture · security · capabilities
·INSTALL / 4 PATHS
/install · cargo · source · docker · plugin
·FEATURED SKILLS · 6 OF 18 · 4 GROUPS
SEE ALL 18 → /new-feature
Orchestrates the seven-subagent feature pipeline with transactional traceability end-to-end. ·PLANNING
/consultancy
Explore problems with full Kvendra context, then persist the findings as typed decisions. ·PLANNING
/bug
Orchestrates the six-subagent bug pipeline with draft and active entities. ·EXECUTION
/validator
Verify changes at three levels: basic, professional, exhaustive. ·QUALITY
/release-manager
Create, manage and close releases with changelog and regression gates. ·RELEASE
/incident-manager
Create incident issues with root-cause analysis and postmortem trail. ·RELEASE
When you'd rather not.
Hosted, managed, ready in 3 minutes.
·WAITLIST
Team
·SHARED WORKSPACES
Soon
REGISTER FOR EARLY ACCESS
- Everything in Pro
- Shared workspaces across the team
- Central audit log
- Skills: required by policy
- CLI broker: required by policy
·POLICY-MANAGED · GOVERNANCE-FIRST
·CONTACT
Enterprise
·SSO · SAML · ON-PREM
Custom
CONTACT SALES
- Everything in Team
- SSO / SAML / SCIM
- On-prem deployment (optional)
- Dedicated infrastructure
- Skills + CLI: required by policy
·POLICY-MANAGED · GOVERNANCE-FIRST
The code is open.
The help is yours if you want it.
·LIVE
Community
·FREE · BEST-EFFORT
·SLA
None community
·SCOPE
Public open
·CHANNELS GitHub Issues · Discord · Docs
- Public bug reports and feature requests
- Community-driven answers from maintainers and users
- Public roadmap and changelog
- No SLA, no guarantees — just the people
·CONTACT
Business
·BESPOKE · NEGOTIATED CASE BY CASE
·SLA
Custom negotiated
·SCOPE
Dedicated slack
·CHANNELS Private Slack channel · Video calls · everything in Community
- Custom SLA, scoped to your stack
- Architecture reviews and migration planning
- On-prem deployment support
- Custom skill development guidance
- Optional 24/7 on-call add-on
·NOTE Support plans are complementary to Cloud — you can self-host with a Business support plan, or run Pro Cloud with Community support. Pricing announced with general availability.
EVERY IDE SPEAKS MCP
Claude Code Cursor Continue