Self-Hosted AI Agent Apps
As of August 2026, AIDiveForge tracks 38 self-hosted ai agent apps. The top three by verified-data score are Nimbus, Xalgorix, and adris.tech. Curated self-hosted ai agent apps tracked by AIDiveForge. Listings are verified against each tool's live website and re-checked regularly.
Last updated July 29, 2026 · 38 tools
Ranked by AIDiveForge's verified-data score: data completeness, verification recency, community rating, and real visitor engagement. How we rank · No tool can pay for placement.
1. Nimbus
Nimbus runs a ReAct planning loop that maps a natural-language request to actual cloud actions: querying live AWS or GCP telemetry, generating infrastructure changes, opening PRs on connected repositories, and updating shared architecture diagrams. Approval gates sit between the agent's plan and execution, so nothing ships without a human sign-off. That model works well for incident diagnosis and routine cost optimizations. Where it strains is on cross-account, deeply custom IAM environments — the agent's tool set reflects the scaffolding its maintainers have wired up, and anything outside that surface area requires you to extend it yourself. Self-hosting via Docker or source install keeps sensitive cloud credentials off third-party infrastructure, which is the primary reason platform teams choose it over a SaaS alternative.
PaidOpen SourceAPISelf-hostedVerified Jul 8, 2026
2. Xalgorix
The core loop is detect, chain, verify: the agent runs reconnaissance through injection through authentication testing, then executes a dedicated validation phase before anything reaches your report. On a public deliberately-vulnerable target, the vendor documents 9 verified findings including a CVSS 9.8 RCE in 17 minutes. The REST API and cron-style scheduling let security teams wire scans directly into CI/CD gates, so releases block on verified findings rather than scanner noise. Where the architecture shows its limits: scan depth and concurrency are credit-gated, and teams running continuous coverage across a wide attack surface will need to budget credits carefully. Self-hosted deployment is listed as an option for teams with data-residency requirements.
PaidOpen Sourcefrom $1 per scanAPISelf-hostedVerified Jul 8, 2026
3. adris.tech
adris is a desktop app for Windows and Linux that bundles eight modules — AI agents, automation, a code editor, local model hosting, a credential vault, DNS-level threat blocking, cross-machine RAM pooling, and a shared knowledge graph — under one login. The agents (called Krew) research prospects and verify contacts in a live browser, then hand results directly to automations that push to Slack, Sheets, or Notion on schedule. Everything stores locally in SQLite; credentials never leave the device. The ceiling appears when you need a public API to connect adris to an existing internal system — the vendor does not list one. Teams that need to pipe agent output into a custom backend will hit that wall fast.
Paidfrom ₹0Self-hostedVerified Jul 20, 2026
4. Lunen.ai
A subject-matter expert describes what they want in plain language; Lunen drafts a structured execution plan with named tools, scoped data, and a schedule — no canvas, no YAML. Every MCP tool connection becomes a per-tool policy decision: allow it to run unattended, or pause for a human sign-off before each call. User actions and agent actions land in the same audit log, which means security reviews have a single trail to pull. The ceiling appears when teams need conditional branching between agent steps — the plain-language plan model does not surface that logic visibly, so complex multi-step dependencies require workarounds the interface does not directly support.
PaidSelf-hostedVerified Jul 20, 2026
5. ami
Ami builds a context graph in SQLite and markdown on disk — tracking entities, relationships, your past decisions, and your writing style — so the agent gets less hand-holdy the more you use it. It maintains a live to-do list and executes recurring busy work by learning how you handle tasks, not by following a static playbook. The self-hosting story is real: no data leaves your machine, no org-level sharing. Where it breaks is scale and surface area — the repo has 3 commits and 4 stars at time of writing, which means production edge cases are yours to discover and debug, not documented anywhere. Teams that need multi-user workflows, audit logs, or a managed API surface will hit the ceiling fast.
FreeOpen SourceSelf-hostedVerified Jul 23, 2026
6. Craft Agents
Built by the craft.do team for their own internal agent work, Craft Agents connects to external services like Slack or Linear, imports skills from Claude Code, and runs local MCP servers as subprocesses — all without requiring CLI fluency from the people reviewing agent output. The document-based interface lets you run multiple agents in parallel and share sessions, which removes the 'only the dev knows what happened' problem. Where it strains: the GitHub issues tracker shows 311 open issues and 146 open pull requests, signaling a project moving fast with rough edges still exposed. Teams expecting enterprise-grade stability at scale will feel those gaps quickly.
FreeOpen SourceSelf-hostedVerified Jul 14, 2026
7. Cybara
Cybara is a fully open-source, MIT-licensed agent runtime you run entirely on your own hardware, with no required account, no telemetry, and no cloud dependency. The vendor describes 50+ model providers, 80+ built-in tools, and 25+ messaging channels available out of the box — a surface area that takes serious effort to configure correctly before it pays off. The operator controls are the genuine differentiator: per-session tool allowlists, filesystem checkpoints, spend caps, and approval gates mean agents don't act without your sign-off. The breadth is also the trap — teams starting with a single use case will spend time disabling or routing around features that don't apply to them. Where it earns its complexity is multi-channel deployments or on-chain automation where the alternative is building that policy layer from scratch.
FreeOpen SourceSelf-hostedVerified Jul 17, 2026
8. Hearth
Hearth runs on your own hardware and handles the tasks that usually demand a SaaS subscription: opening applications, reading and writing files, driving a real browser you can watch, and carrying memory of past sessions — all without a single request leaving your network. The MIT license means you can fork it, extend it, and ship modified versions without legal friction. That said, the GitHub repo shows 9 stars and 297 commits from a single-org project, which signals early-stage software rather than a hardened production runtime. Windows is the primary target; Linux and macOS support is not confirmed by the page. Teams that need cross-platform deployment or enterprise support will hit the ceiling fast.
FreeOpen SourceSelf-hostedVerified Jul 19, 2026
9. Last EHR
The tool wraps four FHIR operations — patient search, chart read, note creation, and observation recording — into a chat agent that runs as the signed-in user, bounded by whatever Medplum AccessPolicy you already have. Every write surfaces as an approval card showing exactly what will be saved before it touches the chart. The same four tools expose as an MCP server for Claude Desktop or Claude Code, read-only by default. The ceiling is visible early: four tools, one backend, one EHR platform. Teams that need more than that are writing their own.
FreeOpen SourceAPISelf-hostedVerified Jul 9, 2026
10. Nova
Nova runs 24 specialist agents on your own machine against your own API keys, storing everything in local SQLite. The pipeline is explicit — classify, decompose, prepare, then stop at a gate before anything publishes, sends, or spends. That gate is the differentiator. Where it strains: the roster of 24 agents with no hosted fallback means setup requires real technical lift, and the Bun + TypeScript stack narrows who can extend it. Teams that outgrow the built-in connectors or need agents to reason across domains the roster doesn't cover will find themselves writing custom system prompts before they expected to.
FreeOpen SourceSelf-hostedVerified Jul 26, 2026
11. Osaurus
Osaurus runs on Apple Silicon via Ollama, MLX, or LM Studio, fully offline — Wi-Fi off, still working. Drop a folder, assign a task, and agents read, write, and execute against your local file system while you're away. When a task outgrows what the local model can handle, you route to ChatGPT, Claude, or Gemini without losing the shared persistent memory thread. The MIT license means no usage caps and no billing — ever. The hard ceiling is macOS exclusivity: teams on Linux or Windows are looking at a different tool from day one.
FreeOpen SourceSelf-hostedVerified Jul 14, 2026
12. Snippbot
Snippbot installs via a single pipx command and runs entirely on your own hardware — no cloud relay, no data leaving your network. The core model is a bench of domain-specific agents (bookkeeping, development, design, project management) each with isolated persistent memory backed by a vector store, full-text search, and a typed knowledge graph. Drop those specialists into a shared chat and they pull from their own episodic memory to collaborate on cross-functional work without context bleed between them. The platform is in open beta and the GitHub repository is listed as coming soon, which means community support and third-party integrations are sparse. Teams that need enterprise connectors or a mature plugin ecosystem will hit that wall quickly.
PaidOpen SourceSelf-hostedVerified Jul 17, 2026
13. Sparkflows
The vendor describes a unified platform covering AI agent building, ML model deployment, no-code data prep, chat assistants, and BI dashboards — all deployable on-premise or across cloud providers. The 50+ pre-built agent templates and 200+ workflow templates mean a data team can reach a working prototype without writing infrastructure glue code. The low-code canvas handles straightforward pipelines well; community reports and the vendor's own positioning toward Alteryx migration suggest it targets teams that have outgrown point solutions. Where it shows strain: complex conditional branching across agents at production scale stretches what a visual canvas can express cleanly, and the free tier is a trial-length access point, not a permanent free seat.
PaidSelf-hostedVerified Jul 26, 2026
14. USB AI Agent
The project ships 13 tools — DuckDuckGo deep search, OSINT via Holehe (121+ sites) and Maigret (600+ platforms), file read/write, Python and shell execution, and persistent memory that saves directly to the drive. Everything runs locally, leaving zero traces on the host machine, which matters on Tails or air-gapped hardware. The autonomous loop lets the agent decide which tools to call and in what order without you directing each step. Where it breaks: the repo has 2 stars and 4 commits, which signals a solo early-stage project with no documented community, no issue history, and no validation of the tool-calling loop at scale or across edge-case inputs.
FreeOpen SourceSelf-hostedVerified Jul 24, 2026
15. Agency Agents
The project is a MIT-licensed, self-hostable collection of pre-defined agent definitions organized by domain — engineering, marketing, product, design, and more — built to be activated inside Claude Code, Cursor, and similar AI coding tools. Each agent carries a defined personality, a stated process, and expected deliverables, so the session opens with role context already loaded. The differentiator is breadth plus specificity: you are not configuring a blank agent; you are picking a specialist with an opinionated approach baked in. The ceiling appears when your workflow requires branching between agents at runtime or dynamic handoffs — the repo defines agents, it does not orchestrate them. Teams needing cross-agent coordination wire that logic themselves on top.
FreeOpen SourceSelf-hostedVerified Jul 4, 2026
16. AutoGPT
The AutoPilot interface lets you describe a workflow in plain language and get a running agent without touching a flow diagram — which means a non-technical PM can ship something before an engineer has opened the docs. When the logic gets complicated, the Builder canvas lets you drag, branch, and loop at the block level, with live input/output inspection at every step. The Dashboard surfaces spend per agent and flags anything waiting on your input, so nothing runs invisible. The Marketplace gives you community-built agents with real run counts, skipping the blank-page problem entirely. Where it strains: teams reporting that complex multi-step branching inside the visual canvas hits readability limits, and the self-hosted path requires managing your own infrastructure when the hosted platform's scale limits bite.
PaidOpen SourceAPISelf-hostedVerified Jul 1, 2026
17. Loma
Loma sits across your tools — Slack, docs, CRM signals — running agents that handle pre-meeting briefs, RFP responses, bug triage, and onboarding health checks without waiting to be asked. The differentiating claim is the context layer: every resolved ticket, closed deal, and fixed bug is stored as a pattern or skill that future agents draw on, so day 100 is meaningfully faster than day 1. Self-hosted under Apache-2.0, it supports Claude, GPT, and Gemini with swap-anytime routing. The vendor states agents complete RFP questionnaires at ~95% coverage, flagging the remainder for human review. Where it strains is in the gaps the scraped content leaves open — enterprise auth, SLA guarantees, and mature operational tooling are not documented.
FreeOpen SourceSelf-hostedVerified Jul 2, 2026
18. OnBuzz
OnBuzz, built by Loxia, lets you spin up multiple autonomous agents that collaborate on tasks, connect directly to LLM providers, and execute work without a cloud intermediary. The Apache-2.0 license and self-hosted design mean your data and your agents stay where you put them. It ships as an Electron app with installers and binaries, so setup does not require hand-rolling a container stack. The tool-use and inter-agent collaboration model is genuinely capable — agents can hand off tasks, run in parallel, and schedule work without you babysitting. Where it strains: the community repository has modest GitHub traction (33 stars at time of indexing), which means documentation gaps surface quickly and community debugging support is thin.
FreeOpen SourceSelf-hostedVerified Jun 27, 2026
19. AnyFrame
AnyFrame lets engineering, ops, and support teams spin up agents that trigger from Slack messages, Linear tickets, or GitHub PR comments and then act — rolling back a deploy, writing tests against a diff, or navigating a billing portal without touching an API. The harness layer is swappable: Claude Code, Codex, Cursor, Gemini CLI, and others sit behind the same agent surface, so a model switch doesn't break your workflow. The SDK lets you embed that same runtime inside your own product in a few lines of code. The ceiling shows up when you need strict approval before an agent acts on production — the vendor describes autonomous execution, and teams that need a mandatory human sign-off step before every consequential action will need to build that gate themselves.
PaidFree tier 500 credits, then pay-as-you-goAPISelf-hostedVerified Jun 9, 2026
20. Autoheal
AI platform leveraging a Production Context Graph to automate alert triage, root cause investigation, and incident remediation for enterprise SRE teams.
PaidAPISelf-hostedVerified May 16, 2026
21. Ciris
CIRIS runs a signed reasoning agent on your phone or a home device, with no warehouse in the middle for the closest privacy circles. The vendor describes two paths: fully on-device using a small model like Gemma 4, or free hosted inference for phones that can't run a local model — both paths produce cryptographically signed outputs. Every claim the agent makes carries an ed25519+post-quantum signature, so you can audit it, revoke trust, and re-open any conclusion built on a bad source. The architecture depends on a 'social circle' data model; data in your innermost circles never sends the network message that would let anyone request it. Teams needing broad third-party integrations or a hosted API endpoint will find neither here.
FreeOpen SourceSelf-hostedVerified Jun 12, 2026
22. Due Diligence Agents
The tool runs parallel analysis across Legal, Finance, Commercial, Technology, Cybersecurity, HR, Tax, Regulatory, and ESG workstreams — domains that siloed consultants hand off sequentially, bleeding weeks in the process. Each agent cross-references findings against the others, so a revenue concentration risk in the commercial workstream gets flagged against the indemnification language in legal without a human manually connecting the dots. Outputs land in Excel and Word with citations intact, ready for an IC memo. The knowledge compounds across deal runs, so repeat buyers in the same sector start with context the first team had to build from scratch. The ceiling appears when your data room contains formats the parser does not handle cleanly — and at that point, teams are pre-processing documents manually before the agents ever see them.
FreeOpen SourceSelf-hostedVerified Jun 1, 2026
23. Extella.AI
The structured tool data describes an agentic execution platform from Chariot Technologies Lab., Inc. with primitives called Rules, Concepts, and Experts — built for research automation, cross-system operations, and persistent memory across sessions. The scraped page, however, describes Spotter: a mobile app that identifies landmarks, street food, and wildlife via camera snap and saves them as travel journal entries. There is no matching factual source to ground a production review of the intended tool. Writing a listing from the validator summary alone, without page-sourced specifics on architecture, failure modes, or integration depth, would produce claims that cannot be verified.
FreeAPISelf-hostedVerified Jun 4, 2026
24. Goose
Goose runs as a desktop app, CLI, or embeddable API — built in Rust, so the performance profile is consistent across macOS, Linux, and Windows without a runtime you have to manage separately. The extension system connects to 70+ tools via the Model Context Protocol, meaning a workflow touching GitHub, Google Drive, and a database isn't stitched together with custom glue code — the standard handles the handoff. Recipes let you capture multi-step workflows as YAML configs and share them across a team or drop them into CI. Where the architecture shows its limits: complex conditional branching inside recipes is not the same as writing that logic in code, and teams building workflows that require dynamic decision trees at depth report dropping into Python extensions to compensate — at which point they are maintaining two systems. Community support is Discord-first; the vendor states no paid tier, so production SLA expectations need to be reset before an org-wide rollout.
FreeOpen SourceAPISelf-hostedVerified Jun 20, 2026
25. Hermes Desktop
Hermes Studio is an open-source, self-hosted dashboard that wraps Hermes Agent in a control plane: task scheduling, multi-agent coordination, memory and skill management, cost tracking, and an approval gate for actions you don't want running unsupervised. The vendor describes it as MIT-licensed with no paid tiers, which means every feature ships without a paywall. The architecture assumes you are already running Hermes Agent locally — Hermes Studio is the interface, not the runtime. Teams that need cloud-hosted infrastructure or agents that run without a local Hermes Agent install will hit that wall immediately.
FreeOpen SourceAPISelf-hostedVerified Jun 9, 2026
26. hermes-desktop
The application handles installation and configuration of Hermes Agent through a desktop interface, which means you're not hand-editing YAML to connect tools or restart sessions. From the same window you manage skills, memory, messaging gateways, and scheduled tasks — the pieces that usually require separate configuration surfaces or CLI literacy. The 278 open issues on the repository signal an active but unpolished project, so expect rough edges on non-standard setups. Teams that need production-grade reliability or multi-user access will hit walls the desktop model was never designed for. This fits a solo developer or small team running a local agent workflow, not an ops team deploying to shared infrastructure.
FreeOpen SourceSelf-hostedVerified Jun 19, 2026
27. Kimi WebBridge
The platform handles long-horizon coding tasks, parallel document research, and full-stack web generation through a coordinated swarm architecture — the vendor states K2.6 scales to 300 sub-agents running concurrently. The model weights are open-source under a Modified MIT license, so teams with strict data governance can run inference locally rather than routing sensitive payloads to a cloud endpoint. Where the friction surfaces is at the edges: the scraped interface shows a broad surface — Slides, Websites, Docs, Deep Research, Sheets, Agent Swarm, Kimi Code, Kimi Claw — and integrating any of those outputs into an existing CI/CD pipeline requires API work the UI does not abstract. Teams building beyond Kimi's native surfaces reach for the API fast.
Paid$19-199/month for subscriptions; $0.95/$4.00 per M tokens for APIAPISelf-hostedVerified Jun 1, 2026
28. Konxios
The core bet is that your agents — code reviewer, personal assistant, browser automator — live on your machine, talk to each other, and never push your data to a third-party server. Local models run through Ollama or LM Studio; cloud fallback goes through OpenAI, Anthropic, or OpenRouter when you need it. Docker isolation means each project gets its own sandboxed container, so a misfired agent cannot touch unrelated work. The platform is in public beta at v0.1.0, which means the agent skill marketplace, multi-agent collaboration depth, and edge-case reliability are still being shaped by early users — not by two years of production hardening. Teams that need proven uptime SLAs or audit trails for enterprise compliance will hit the beta ceiling fast.
PaidSelf-hostedVerified Jun 20, 2026
29. LobeHub
LobeHub lets you define a goal and have the system assemble an agent team, dispatch parallel workers across tasks, and surface results without you approving every step. The agent marketplace and skill library — reportedly over 332,000 skills and 64,000 MCP server connections — mean you're not building from scratch each time. Memory is white-box and editable, so agents don't silently drift from your preferences. Where it gets difficult: the self-hosted path requires you to manage your own infrastructure, and the complexity of multi-agent coordination means debugging a failed task chain is non-trivial. Teams running production workloads tend to add observability tooling — the Langfuse integration listed on the page suggests this is an expected pattern, not an edge case.
Paid$9.9/moAPISelf-hostedVerified Jun 9, 2026
30. Locaible
Locaible runs AI agents entirely on your own machine: no bytes leave the device, no API calls to OpenAI or Anthropic, no telemetry. The vendor states it is GDPR and EU AI Act compliant by design, which matters when your legal or finance team needs a paper trail for the regulator, not a ToS URL. Multi-step workflows chain separate agents — one retrieves from your indexed documents, one analyses, one drafts — each running its own local model. The ceiling appears when your team scales beyond a small LAN setup: team seats authenticate over a private token and require a detected LAN IP, so distributed or remote teams hit a networking configuration wall before they hit a workflow one.
PaidFree Trial · 7 daysAPISelf-hostedVerified Jun 10, 2026
31. MagesticAI
The platform runs a pipeline of specialized agents — Planner, Coder, QA — that hand off work through isolated Git worktrees, so each task gets its own branch and a bad run does not contaminate the main codebase. You monitor execution in real-time through a web UI, which means you are not staring at terminal logs hoping the right thing happened. The vendor describes cross-session knowledge retention, so the system carries context between separate task runs. The architecture supports multiple LLM providers, which means you are not locked to one API when costs shift. At 78 stars and 184 commits, this is early-stage software — community support is thin and the blast radius of an undocumented breaking change falls entirely on your team.
FreeOpen SourceSelf-hostedVerified Jun 9, 2026
32. NanoClaw
NanoClaw is a lightweight, open-source personal AI agent that runs on your own machine, connects to messaging apps like WhatsApp, Telegram, Slack, Discord, and Signal, and is built around just 15 source files you can read in a single sitting.
FreeAPISelf-hostedVerified Apr 23, 2026
33. OpenLegion
Each agent gets its own isolated container, spend cap, and vault-proxied credentials — so a rogue agent can't drain your API budget or leak credentials to the next task in the queue. The platform deploys a coordinated fleet from a plain-English description of the function you need: a sales pipeline, a content studio, a research desk. Credential handling and per-agent budgets are locked down by default, which means you're not retrofitting security after something goes wrong. The ceiling appears when your workflow needs branching logic that the template model can't express — at that point you're describing edge cases in natural language and hoping the agent interprets them correctly. Teams with deterministic multi-step requirements often add a separate orchestration layer to compensate.
PaidFree Trial · 7 days$19/moAPISelf-hostedVerified Jun 1, 2026
34. Orchestrik.ai
The scraped vendor page does not match the tool data provided. The page content describes 'Spotter,' a travel-identification app, while the structured data references an enterprise AI agent platform from ITMTB Technologies. Because the only factual source available is the Spotter page — which contains no information about multi-agent workflows, compliance features, audit trails, or backend integrations — this listing cannot be written to the publication standard required. Asserting capabilities from the structured input without page-level sourcing would violate the grounding rule. A corrected scrape of the ITMTB Technologies product page is needed before this listing can be completed accurately.
Paid₹5,000–₹12,500/month base + usage overagesAPISelf-hostedVerified Jun 2, 2026
35. SynapCores Agent
The repo, published by SynapCores under MIT, routes all memory, retrieval, semantic tool selection, and generation through the SynapCores backend — one database as the entire brain. There is no LangChain, no separate vector store, no framework glue to audit or upgrade. The project ships a browser chat widget and a live debug sidebar so you can watch memory recall and tool routing decisions in real time. That transparency is the differentiating feature — and also the boundary: the agent's intelligence rides entirely on the SynapCores backend, whose self-hosted deployment requirements the repo does not fully document. Teams that need the backend running on-premise will hit that wall before they hit a code problem.
FreeOpen SourceAPISelf-hostedVerified Jun 9, 2026
36. Teralynk
The scraped page content does not match the tool described in the structured data — the page belongs to Spotter, a travel identification app, not Teralynk's workflow automation platform. No production details about Teralynk's agent architecture, file system integrations, MCP tool use, or governance controls can be sourced from the provided page. The vendor states a freemium model with storage limits and capped workflow runs on the free tier; paid-only features unlock higher run volumes and expanded storage. Teams evaluating this for compliance auditing or multi-cloud document workflows cannot rely on this listing for verified capability claims — vendor documentation should be consulted directly.
Paid$9.99/moAPISelf-hostedVerified Jun 1, 2026
37. Vibesolve
The tool takes a plain-English description of a scheduling, routing, or assignment problem and runs it through a multi-agent pipeline that extracts a structured spec, generates constraint code, and packages everything in a Docker container you can stand up with a single command. Before code is generated, you review and correct the extracted spec in plain English — which means misunderstood requirements surface before they cost you a debugging session. The self-healing loop catches compile errors automatically, so you are not hand-tuning syntax. The vendor states this is an experimental tool built for prototyping, not production; the consulting arm exists precisely because the gap between a generated prototype and a production-grade solver is real and often large.
PaidAPISelf-hostedVerified Jun 25, 2026
38. WorkBuddy
WorkBuddy runs as a local-first agent on the desktop, autonomously chaining file access, web search, and document generation into single-prompt workflows. The Tencent ecosystem fit is real: WeCom and WeChat integrations mean scheduling and messaging tasks route without extra setup, which matters if your organization already lives there. Outside that ecosystem, the integration surface narrows fast. Teams running mixed SaaS stacks report reaching for MCP-compatible connectors to fill the gaps — which adds configuration overhead the tool is supposed to eliminate. Self-hosted execution is the headline privacy story, but the closed-source codebase means you audit what the vendor discloses, not the code itself.
Paid$9.95/moAPISelf-hostedVerified Jun 5, 2026
Listings on this page are sourced and verified by the AIDiveForge data pipeline. AIDiveForge is editorially independent — no money changes hands for inclusion.