AgentIndex icon
AgentIndex
ToolsMCP ServersHOTCategoriesTrendingNewBest ListsCompareGuides
Submit Tool
ToolsMCP ServersHOTCategoriesTrendingNewBest ListsCompareGuides
Home/
Multi-Agent/
marvis
marvis logo

marvis

Active·★ 14·NOASSERTION·Updated 2026-09-22

Marvis describes itself as a company brain for coding agents. It captures what your agents do as they work (decisions, specs, learnings, mistakes), organises it into a cross-project knowledge graph, and hands it back to them, so Claude Code, Codex, Cursor or Gemini sessions do not start from zero. A reflective layer compares agent output with your stated intent and flags drift early.

The README lists a graph with 17 deterministic edge types, an append-only audit log, a five-layer reflection pipeline, rule gates enforced by hooks, and an MCP server with 91 tools across 11 categories, including tasks, handoffs, semantic search, costs, pull requests and graph queries. Search blends a self-hosted multilingual embedding model with SQLite BM25. Install is a Claude Code plugin or `pip install marvisx-cli`, and a Docker Compose template runs the full console and API stack. A paid MarvisX tier adds managed hosting, multi-department teams and EU AI Act logging.

This write-up is based on the project's README and public repository data. We have not hand-tested the tool. Data as of 2026-10-17.

License
Not detected by GitHub
Latest release
v0.4.8 · 2026-08-31
Last commit
2026-09-22
Contributors
6
Forks
1
Open issues and PRs
10
First published
2026-05-25
PyPI package
marvisx-cli 0.4.8 · 20 downloads last week

Source: GitHub, npm and PyPI public APIs.

#AI Agents#Knowledge Graph#Project Management#Workflow Automation#Self-hosted#LLM Tooling#Agent Coordination#Audit Log
$ Install
$ pip install marvisx-cli && marvis init
↗ Visit site★ GitHub
01

Features

01Agent memory persistence through captured work artifacts.
02Cross-project Knowledge Graph for linking decisions, code, and documents.
03Constitution-enforced safety rules and an immutable audit log.
04Agent-native Management and Control Plane (MCP) surface for seamless integration.
02

Why choose it

+Gives agents persistent, cross-project memory captured as a side effect of normal work, not retyped by hand.
+Append-only audit log and rule gates enforced by hooks suit teams that need traceability.
+Wide MCP surface (91 tools) works from Claude Code, Codex, Gemini and OpenCode, with parity with the web console.
+Can be fully self-hosted, with a local SQLite mode for a single user.
03

Trade-offs

!The code is under the Business Source License 1.1, which is source-available and not OSI open source; it converts to Apache 2.0 four years after release, and hosting it for third parties needs a paid licence.
!The README admits the default self-hosted embeddings give lower retrieval quality than commercial managed providers.
!The full feature set is large and the project is new with about 14 stars; some capabilities belong to the paid MarvisX tier.
04

Maintenance check

ActiveChecked: 2026-10-19
Last commit
2026-09-22
Commits, last 90 days
100+
Releases, last 90 days
1
Latest release
v0.4.8 (2026-08-31)
Open issues
10
Recent issues with at least one comment
3/14
Commits per week, last 12 weeks
Commits per week, last 12 weeks: 0, 0, 0, 34, 23, 1, 0, 2, 0, 0, 0, 0

Oldest week on the left; the latest four weeks are orange. Source: GitHub commit activity.

Status rule: archived on GitHub = Archived; last commit within 90 days = Active; 91 to 365 days = Slowing; older = Stale. All figures come from the GitHub API on the date shown.

Most discussed open issues

  • #38 fix(kg): local brains have no cross-project graph connectivity (Universe shows isolated project dots, degree never computed)1 comments · 2026-06-19
  • #40 fix(console): la sidebar Progetti non si aggiorna dopo il seed demo del wizard (serve reload)0 comments · 2026-06-19
  • #39 fix(console): pagina Settings -> Activity (audit) chiama /api/v1/audit-log inesistente -> 4040 comments · 2026-06-19
  • #36 BYOK: 'Aggiungi chiave' disabilitato sul tier locale — BYOK_FERNET_SECRET non auto-provisionato0 comments · 2026-06-17
  • #35 Universe/Cosmo grafo vuoto coi progetti locali — registrazione non crea nodi KG0 comments · 2026-06-17

Issue titles are quoted from GitHub. Open the link to check whether a problem still applies to the version you use.

05

Compatibility

Python
Runtime
Verified via docs
Docker
Deployment
Verified via docs
WSL2
Environment
Verified via docs
Linux
OS
Verified via docs
06

Quick start

1
$ pip install marvisx-cli
2
$ marvis init
07

Use cases

↳AI agents managing and coordinating development projects.
↳Establishing a persistent memory and learning system for agents' activities.
↳Ensuring project alignment and detecting deviation from intent.
↳Providing a governed and auditable environment for agent operations.
08

Alternatives

⚡Top VS:
VSmcp★11→VSjavis-os★279→VSKronn★18→VSCotal★335→
mcp logo
mcp★ 11
Connect your agents to every app through a single MCP.
VSCompare→
javis-os logo
javis-os★ 279
No description provided.
VSCompare→
Kronn logo
Kronn★ 18
Self-hosted control plane for AI coding agents (Claude Code, Codex, Gemini, Ollama). Shared MCPs, reusable workflows, AI audit that makes any repo AI-friendly. Engineering, not prompting : smaller prompts, fewer hallucinations, lower token bill.
VSCompare→
Cotal logo
Cotal★ 335
The open standard for agent coordination
VSCompare→
oto-dock logo
oto-dock★ 193
The self-hosted platform for your company's AI agents. Your team builds them on Claude Code, Codex or local models, and they work in chats, on schedules, on the phone and on your own machines.
VSCompare→
Hexis logo
Hexis★ 117
Centrally manage AI-agent skills, tools, context, permissions, and identity across your entire org. Self-hosted, Git-backed and MCP-native.
VSCompare→
okto-pulse logo
okto-pulse★ 51
Local-first, spec-driven SDLC workbench with governance gates, knowledge graphs, and MCP for AI coding agents.
VSCompare→
OctoForge logo
OctoForge★ 22
Self-hosted multi-tenant agent platform for teams: skills, knowledge, API contracts and MCP servers live in the database and grow at runtime; big API responses are queried with SQL, not stuffed into context. Typed Python core (Protocol ports) you can embed in your own stack.
VSCompare→
See all alternatives →

Comments

Log in to leave a comment

No comments yet. Be the first!

On this page
01Features02Why choose it03Trade-offs04Maintenance check05Compatibility06Quick start07Use cases08Alternatives
emiliomartucci /
marvis
↗

Self-hosted knowledge graph and audit log that agents keep in sync as they work

141
Last commit1mo ago
StatusActive
CategoryMulti-Agent
Contributors6
Latest releasev0.4.8
Links
Documentation↗Discussion↗Issues↗Releases↗
Cloud Infrastructure
Ad

Deploy on DigitalOcean

Host your MCP servers, Docker containers, and AI agents with $200 free credit over 60 days.

Claim $200 Free Credit
© 2026 AgentIndex.app|Built by a 10-year iOS Developer.
Best by use caseAboutContactAdvertisePrivacy PolicyTerms of UseQYSGitHubBuy me a coffee ☕

Browse by Category

MCP ServersCode AssistantWorkflow AutomationRAG / Knowledge BaseMulti-AgentBrowser AutomationLLM InfraDev ToolingObservability

Not affiliated with Anthropic, OpenAI or Microsoft.