FleetQ

Most agent frameworks give you a Python notebook. FleetQ gives you aproduction platform.

DashboardKPI overview with active experiments, success rate, budget spend, and pending approvals.

Agent Template GalleryBrowse 14 pre-built agent templates across 5 categories. Search, filter by category, and deploy with one click.

Agent LLM ConfigurationPer-agent provider and model selection with fallback chains. Supports Anthropic, OpenAI, Google, and local agents.

Agent EvolutionAI-driven agent self-improvement. Analyze execution history, propose personality and config changes, and apply with one click.

Crew ExecutionLive progress tracking during multi-agent crew execution. Each task shows its assigned skill, provider, and elapsed time.

Task OutputExpand any completed task to inspect the AI-generated output, including structured JSON responses.

Visual Workflow BuilderDAG-based workflow editor with conditional branching, human tasks, switch nodes, and dynamic forks.

Tool ManagementManage MCP servers, built-in tools, and external integrations with risk classification and per-agent assignment.

AI Assistant SidebarContext-aware AI chat embedded in every page with 28 built-in tools for querying and managing the platform.

Experiment DetailFull experiment lifecycle view with timeline, tasks, transitions, artifacts, metrics, and outbound delivery.

Settings & WebhooksGlobal platform settings, AI provider keys (BYOK), outbound connectors, and webhook configuration.

Error HandlingFailed tasks display detailed error information including provider, error type, and request IDs for debugging.

- AI Agents— role, goal, backstory, personality traits, skill assignments, per-agent provider/model fallback chains
- Agent Templates— 14 pre-built templates across 5 categories (engineering, content, business, design, research)
- Agent Evolution— LLM analyzes execution history, proposes config changes, one-click approval
- Agent Crews— Multi-agent teams with coordinator/QA/worker roles, 7 process types (sequential, parallel, hierarchical, self-claim, adversarial, fanout, chat-room), weighted QA scoring
- Pre-Execution Scout Phase— cheap LLM pre-call identifies what knowledge the agent needs → targeted semantic search instead of generic recall
- Step Budget Awareness— agent system prompt targets 80% of allowed steps for core work, reserves the rest for synthesis
- Experiment Pipeline— 20-state machine with automatic stage progression (scoring → planning → building → approval → executing → metrics → evaluating)
- Visual Workflow DAG— 8 node types (agent, conditional, human-task, switch, dynamic-fork, do-while, compensation, sub-workflow). Pre-built Web Dev Cycle template. NL → workflow generator.
- Projects— one-shot and continuous projects with cron scheduling, budget caps, milestones, overlap policies

- BYOK— bring your own keys for Anthropic (Claude), OpenAI (GPT-4o), Google (Gemini)
- Local LLMs— Ollama, LM Studio, vLLM, llama.cpp via OpenAI-compatible endpoints; 17 preset Ollama models; SSRF protection
- Local Agents— Codex and Claude Code as execution backends (auto-detected, zero cost)
- Portkey Gateway— optional drop-in that unlocks 250+ LLM providers with semantic caching and fallbacks
- RunPod GPU Integration— invoke RunPod serverless endpoints or manage full GPU pod lifecycles as skills; BYOK API key; spot pricing
- Pluggable Compute Providers—gpu_computeskills backed by RunPod, Replicate, Fal.ai, Vast.ai
- AI Gateway— provider-agnostic via PrismPHP with 6-layer middleware (rate-limit, budget, idempotency, semantic-cache, schema-validation, usage-tracking), circuit breakers, fallback chains
- Semantic Cache— pgvector-backed cosine similarity (threshold 0.92) cross-team cache — cuts LLM spend on repeat prompts

- Signal connectors— 20+ drivers: webhook, RSS, IMAP, Slack, Discord, WhatsApp, GitHub, Linear, Jira, PagerDuty, Sentry, Datadog, ClearCue, Telegram, Matrix, Notion, Confluence, Screenpipe, Searxng, more
- Bug Report signals— lightweight QA pipeline with public JS widget, screenshot + console + network + action log capture, threaded comments (reporter + agent + support), agent delegation, SLA escalation
- Trigger rules— event-driven automation with condition evaluator, dry-run testing
- Multi-Channel Outbound— Email (SMTP), Webhook, ntfy plus eight chat channels as first-class drivers (Telegram, Slack, Discord, Microsoft Teams, Google Chat, Matrix, Signal, Supabase Realtime), each with a config page, rate limiting and blacklist
- Webhooks— inbound (HMAC-SHA256) + outbound (retry, event filtering)

- Approvals— inbox with SLA enforcement + escalation
- Human Tasks— embedded form schemas on workflow nodes
- Credit Ledger— per-experiment and per-project with pessimistic locking and auto-pause on overspend
- Credential Vault— encrypted external service credentials with rotation, OAuth2, expiry tracking, per-project injection
- SSH tools— TOFU (Trust On First Use) fingerprint verification, per-tool allowed-commands whitelist, multi-layer command security policy
- Audit Trail— full activity log (spatie/activitylog), searchable + filterable
- Tenant Isolation— multi-layerTeamScope+BelongsToTeam+withoutGlobalScopes()discipline

- Integrations— GitHub, Slack, Notion, Airtable, Linear, Stripe, Vercel, Netlify, generic webhook/polling with OAuth 2.0
- Autonomous Web Dev Pipeline— agents can open PRs, merge, dispatch CI workflows, create releases, trigger Vercel/Netlify/SSH deploys through MCP tools
- Website Builder— AI-generated static sites with 8 widget types, Vercel + ZIP deployment drivers, form submissions, blog/navigation/contact widgets
- Founder Mode pack— marketplace bundle of 6 persona agents (Strategist, Product Lead, Growth Hacker, Finance Advisor, Ops Manager, Risk Officer), 20 framework skills (RICE, SPIN, BANT, MEDDIC, OKRs, Shape Up, Unit Economics, Kano, TAM-SAM-SOM, K-Factor, NPV-IRR, RACI, A/B Testing, OWASP), 5 pre-built workflows
- Marketplace— browse, publish, install shared skills, agents, workflows, and bundles with AI risk scanning

- REST API— 175+ endpoints under/api/v1/with Sanctum auth, cursor pagination, auto-generated OpenAPI 3.1 at/docs/api
- MCP Server—675+ Model Context Protocol tools across 45 domains (62 tool groups)(stdio + HTTP/SSE + OAuth2/PKCE)
- Real-World Action governance—ActionProposalflow gates assistant tool calls, integration writes, and git pushes through a per-tier risk policy with auto-execute on approval
- Public discovery endpoint—GET /.well-known/fleetqreturns a config-gated capability manifest so external AI tools can auto-configure
- Live team graph—/team-graphpage with real-time updates via Laravel Reverb WebSockets
- Structured MCP errors— canonical gRPC-style error codes (UNAVAILABLE,PERMISSION_DENIED,RESOURCE_EXHAUSTED,DEADLINE_EXCEEDED,INVALID_ARGUMENT,FAILED_PRECONDITION,NOT_FOUND,INTERNAL) with retryable hints — agents know when to retry vs. fail fast
- Per-tool deadlines— optionaldeadline_msparameter on every MCP tool; agents can bound wall-clock time per call
- OpenTelemetry tracing— OTLP HTTP exporter, Jaeger all-in-one viadocker compose --profile observability up, spans for MCP tool → AI gateway → LLM provider
- Tool Management— MCP servers (stdio/HTTP), built-in tools (bash/filesystem/browser), risk classification, per-agent assignment
- MCP client compatibility— Claude Desktop, Claude.ai, ChatGPT Apps, Cursor, Codex, Claude Code, Gemini CLI, any OAuth2 client

- Queue Management— Laravel Horizon with 6 priority queues and auto-scaling
- Testing— regression test suites for agent outputs with automated evaluation
- Per-Call Working Directory— local/bridge agents can operate in a configured working directory per-agent, isolated project contexts

FleetQ is built for teams running AI agents in production, not toy demos.

- Autonomous dev pipelines— agent opens PR → CI runs → reviewer agent approves → merge → deploy. Human approves only on risk signals.
- Customer support triage— bug report widget → agent extracts reproduction steps from console/network log → experiment runs → notifies reporter with fix or agent-generated workaround.
- Multi-agent research— crew of Strategist + Researcher + Writer with QA reviewer. Each step weighted by domain rubric.
- Scheduled content ops— continuous project runs daily, each run executes a DAG: draft → review → SEO-check → publish → schedule social.
- Incident response— PagerDuty/Sentry signal → trigger rule → diagnosis agent → human approval on runbook action → Slack notify.
- GPU workloads— agent callsgpu_computeskill on RunPod serverless (Whisper, FLUX, Bark) as part of a larger workflow, with cost accounting.
- Local-first agent dev— Ollama + Codex + Claude Code auto-detected, zero API cost for prototyping; switch to cloud providers for production.
- Bring FleetQ into Claude— expose your internal data + tools as MCP server, Claude Desktop/ChatGPT/Cursor can drive the platform programmatically.

TL;DR — if you're building production agent systems with LLMs and want visual workflows + MCP + human oversight, FleetQ is the only platform that bundles all of it.

git clone https://github.com/escapeboy/agent-fleet-o.git cd agent-fleet make install

- Copy.env.exampleto.env
- Build and start all Docker services
- Run the interactive setup wizard (database, admin account, LLM provider)

Requirements: PHP 8.4+, PostgreSQL 17+, Redis 7+, Node.js 20+, Composer

git clone https://github.com/escapeboy/agent-fleet-o.git cd agent-fleet composer install npm install && npm run build cp .env.example .env # Edit .env — set DB_HOST, DB_DATABASE, DB_USERNAME, DB_PASSWORD, REDIS_HOST php artisan key:generate php artisan migrate php artisan horizon & php artisan serve

Alternative:Runphp artisan app:installfor an interactive CLI setup wizard that also seeds default agents and skills.

- No email verification— the self-hosted edition skips email verification entirely. Accounts are active immediately on registration.
- Single user— all registered users join the default workspace automatically.

If you're running FleetQ locally on your own machine and don't want to enter a password on every visit, setAPP_AUTH_BYPASS=truein.env:

APP_AUTH_BYPASS=true # Auto-login as first user APP_ENV=local # Required — bypass is disabled in production

With bypass enabled, the app logs you in automatically on every request. A logout link is still shown but you'll be logged back in on the next page load — this is intentional.

Warning:Never setAPP_AUTH_BYPASS=trueon a server accessible from the internet.

All configuration is in.env. Key variables:

# Database (PostgreSQL required) DB_CONNECTION=pgsql DB_HOST=postgres DB_DATABASE=agent_fleet # Redis (queues, cache, sessions, locks) REDIS_HOST=redis REDIS_DB=0 # Queues REDIS_CACHE_DB=1 # Cache REDIS_LOCK_DB=2 # Locks # LLM Providers -- at least one required for AI features ANTHROPIC_API_KEY= OPENAI_API_KEY= GOOGLE_AI_API_KEY= # Auth bypass -- local no-password mode (never use in production) APP_AUTH_BYPASS=false

Additional LLM keys can be configured inSettings > AI Provider Keysafter login.

To use local models (Ollama, LM Studio, vLLM):

LOCAL_LLM_ENABLED=true LOCAL_LLM_SSRF_PROTECTION=false # set false if Ollama is on a LAN IP (192.168.x.x) LOCAL_LLM_TIMEOUT=180

Then configure endpoints inSettings > Local LLM Endpoints.

Agents can execute commands on the host machine (or any remote server) via SSH using the built-in SSH tool type. This is useful for running local scripts, interacting with the filesystem, or orchestrating host-level processes from an agent.
- The platform stores SSH private keys encrypted in the Credential vault.
- An SSH Tool is configured withhost,port,username,credential_id, and an optionalallowed_commandswhitelist.
- On the first connection to a host, the server's public key fingerprint is stored viaTOFU(Trust On First Use). Subsequent connections verify the fingerprint — a mismatch raises an error to prevent MITM attacks.
- Manage trusted fingerprints viaSettings > SSH Fingerprintsor thetool_ssh_fingerprintsMCP tool.

Setup (Docker — connecting container to host)

The containers reach the host machine viahost.docker.internal, which is pre-configured indocker-compose.ymlviaextra_hosts: host.docker.internal:host-gateway.

ssh-keygen -t ed25519 -C "fleetq-agent@local" -f ~/.ssh/fleetq_agent_key -N ""

cat ~/.ssh/fleetq_agent_key.pub >> ~/.ssh/authorized_keys chmod 600 ~/.ssh/authorized_keys

Navigate toCredentials → New Credential:

- Type:SSH Key
- Paste the contents of~/.ssh/fleetq_agent_key(private key)

curl -X POST http://localhost:8080/api/v1/credentials \ -H "Authorization: Bearer $TOKEN" \ -H "Content-Type: application/json" \ -d '{ "name": "Host SSH Key", "credential_type": "ssh_key", "secret_data": {"private_key": "<contents of fleetq_agent_key>"} }'

Navigate toTools → New Tool → Built-in → SSH Remote, or via API:

If FleetQ saves you time, a ⭐ helps others find it. GitHub ranks repos by star velocity.