An AI agent orchestration platform with RAG, LangGraph workflows, observability, and evaluation. Manage a roster of agent personas, run them against real infrastructure via SSH, and measure their performance over time.
Agents Platform is a full-stack application for building, managing, and running AI agents. Each agent has a persona, system prompt, skill set, tool inventory, and knowledge sources. The platform dispatches agents via SSH to a remote Claude Code session, tracks every run with cost and latency telemetry, and surfaces the results through a dark glassmorphism dashboard.
It ships with 20 pre-built agent personas covering infrastructure, development, security, media, and automation domains. A demo mode runs locally with docker-compose -- no SSH target needed.
- 20 persona definitions: name, title, tagline, system prompt, skills, tools, MCP servers, knowledge sources, example tasks, and related agents
- Full CRUD via REST API and in-app forms
- Category filtering (infrastructure, development, security, media, automation)
- Per-agent accent color and SVG avatar (20 unique illustrations)
- Runs
claude -p "<prompt>"on a remote host over SSH - Three composition modes: parallel (fan-out, aggregate), sequential (pipeline), meeting (structured debate)
- Cron scheduling with configurable concurrency limits
- Per-run Discord webhook notifications (success and failure)
- Run history with stdout capture and status tracking
- Vector store: Qdrant with Ollama embeddings (
nomic-embed-text) - Pluggable document loaders: Markdown files, YAML, Terraform, URLs, transcripts
- RAG Playground UI: load documents, query the index, inspect retrieved chunks
- LangChain retrieval chain with Anthropic Claude for generation
- Task router: Claude Haiku classifies each request (RAG query / workflow / SSH dispatch)
- State machine graphs built with
@langchain/langgraph - Built-in tools:
kubectlrunner, file reader, RAG search - Workflows UI for inspecting graph state and step-by-step traces
- Telemetry middleware captures every API call: model, tokens in/out, latency, cost
- Cost calculator using Claude model pricing tiers
- Recharts dashboard: daily cost trends, model distribution, latency percentiles, recent trace table
- Eval suite: write test cases, run LLM-as-judge scoring, compare models, detect regressions
The UI is a dark, glassmorphism dashboard.
The 20-agent roster with search, category filters, and quick-task cards.
Ten production-grade scheduled workflows spanning all three composition modes — parallel, sequential, and meeting.
Run history with status, duration, and per-run summaries.
Route visualization for multi-step agent workflows.
Browser (React 18 + Vite)
|
| REST / JSON
v
Express.js (port 3001)
|-- /api/agents Agent CRUD
|-- /api/schedules Cron scheduler
|-- /api/runs Run history
|-- /api/rag RAG ingest + query
|-- /api/workflows LangGraph execution
|-- /api/observability Telemetry traces
|-- /api/eval Evaluation suites
|
|-- SQLite (better-sqlite3, WAL)
| agents, runs, schedules, traces, eval results
|
|-- LangGraph
| Task router (Haiku) --> RAG chain | Workflow graph | SSH dispatch
|
|-- Qdrant (vector store)
| Ollama embeddings (nomic-embed-text)
|
|-- SSH --> Remote Host
claude -p "<system prompt + task>"
(Claude Code CLI, parallel / sequential / meeting modes)
The platform ships with 10 ready-to-use scheduled workflows that exercise all three composition modes against real platform operations — the kind a platform team runs on a cron cadence. Each bundles a curated set of agents, a rich task prompt, and a realistic schedule.
| Schedule | Mode | Cadence | Agents |
|---|---|---|---|
| Nightly Infrastructure Audit | parallel | daily 02:00 | Atlas, Sentinel, Bastion, Patch |
| Security & Compliance Sweep | sequential | Mon 03:00 | Vault, Cipher, Sentinel, Relay |
| Incident Response Drill | meeting | Fri 14:00 | Atlas, Mirror, Bastion, Sentinel, Relay |
| Release Readiness Pipeline | sequential | weekdays 09:00 | Tempo, Dock, Flux, Proxy |
| Cost & Performance Review | parallel | Mon 08:00 | Scout, Sentinel, Oracle, Ledger |
| Backup Restore Verification Drill | sequential | Tue 04:17 | Bastion, Mirror, Ledger, Relay |
| Expiry & Capacity Forecast | parallel | Thu 07:23 | Cipher, Proxy, Atlas, Sentinel |
| Dependency & CVE Patch Triage | sequential | Wed 05:47 | Dock, Patch, Vault, Flux |
| Observability Coverage Audit | meeting | Wed 13:47 | Sentinel, Scout, Relay, Oracle |
| Data Pipeline & Ingestion Health Check | parallel | daily 06:17 | Scout, Oracle, Sentinel, Relay |
Requirements: Docker, Docker Compose, and (for SSH dispatch) a remote host running Claude Code.
git clone https://github.com/kernelpanic09/agents-platform.git
cd agents-platform
cp .env.example .env
# Edit .env -- set ANTHROPIC_API_KEY at minimum
docker compose upOpen http://localhost:3001.
Demo mode is on by default (DEMO_MODE=true in docker-compose.yml). SSH dispatch is disabled in demo mode -- all other features work.
On first start, Ollama needs to pull the embedding model:
docker compose exec ollama ollama pull nomic-embed-text| Variable | Default | Description |
|---|---|---|
PORT |
3001 |
Express server port |
DATA_DIR |
. |
Directory for agents.db SQLite file |
ANTHROPIC_API_KEY |
(required) | Anthropic API key for RAG chat and eval |
QDRANT_URL |
http://localhost:6333 |
Qdrant vector store URL |
OLLAMA_URL |
http://localhost:11434 |
Ollama embedding server URL |
EMBED_MODEL |
nomic-embed-text |
Ollama model for embeddings |
SSH_TARGET |
(required for dispatch) | Remote host in user@host format |
SSH_KEY_PATH |
(optional) | Path to SSH private key |
CLAUDE_MODEL |
sonnet |
Claude model to use for SSH dispatch |
ENABLE_SCHEDULER |
false |
Enable cron scheduler and manual /run endpoint |
MAX_CONCURRENT_RUNS |
2 |
Max parallel SSH dispatch jobs |
DISCORD_WEBHOOK_URL |
(optional) | Discord webhook for run notifications |
DEMO_MODE |
false |
Seed demo data and disable SSH |
| Layer | Technology |
|---|---|
| Frontend | React 18, Vite 5, Tailwind CSS 3, React Router v6 |
| UI Components | Lucide React, Recharts, custom SVG avatars |
| Backend | Node.js, Express.js |
| Database | SQLite via better-sqlite3 (WAL mode) |
| AI Orchestration | LangChain, LangGraph, Anthropic SDK |
| Vector Store | Qdrant |
| Embeddings | Ollama (nomic-embed-text) |
| SSH Dispatch | Native Node.js child_process over SSH |
| Scheduling | node-cron |
| Schema Validation | Zod |
| Containerization | Docker, Docker Compose |
| Method | Path | Description |
|---|---|---|
GET |
/api/agents |
List all agents (excludes system_prompt) |
GET |
/api/agents/:id |
Full agent detail with system_prompt |
POST |
/api/agents |
Create agent |
PUT |
/api/agents/:id |
Update agent (partial) |
DELETE |
/api/agents/:id |
Delete agent |
GET |
/api/agents/:id/prompt |
Raw prompt file content (dev only) |
| Method | Path | Description |
|---|---|---|
GET |
/api/schedules |
List all schedules |
POST |
/api/schedules |
Create schedule (cron + agent + task) |
PUT |
/api/schedules/:id |
Update schedule |
DELETE |
/api/schedules/:id |
Delete schedule |
POST |
/api/schedules/:id/run |
Trigger schedule manually |
GET |
/api/runs |
List all runs with status |
GET |
/api/runs/:id |
Run detail with stdout |
| Method | Path | Description |
|---|---|---|
GET |
/api/rag/health |
Qdrant + Ollama connectivity check |
POST |
/api/rag/ingest |
Ingest document into Qdrant |
POST |
/api/rag/search |
Semantic search across ingested documents |
POST |
/api/rag/chat |
RAG-augmented chat (retrieve + generate) |
GET |
/api/rag/sources |
List ingested document sources |
DELETE |
/api/rag/sources/:id |
Remove source and its vectors |
| Method | Path | Description |
|---|---|---|
GET |
/api/workflows/types |
List available workflow types |
POST |
/api/workflows/route |
Classify a task (RAG / workflow / SSH) |
| Method | Path | Description |
|---|---|---|
GET |
/api/observability/traces |
Recent telemetry traces |
GET |
/api/observability/costs |
Aggregated cost stats (by model, agent, day) |
GET |
/api/observability/latency |
Latency percentiles per model |
| Method | Path | Description |
|---|---|---|
GET |
/api/eval/suites |
List eval suites with case/run counts |
POST |
/api/eval/suites |
Create eval suite |
POST |
/api/eval/suites/:id/cases |
Add test case to suite |
POST |
/api/eval/suites/:id/run |
Run eval suite against a model |
GET |
/api/eval/runs |
List eval runs |
GET |
/api/eval/runs/:id/results |
Per-case results with judge scores |
| Method | Path | Description |
|---|---|---|
GET |
/api/mcp-servers |
List available MCP servers |
GET |
/api/mcp-servers/:id |
MCP server detail |
POST |
/api/mcp-servers/config |
Generate MCP config JSON for selected servers |
| Method | Path | Description |
|---|---|---|
GET |
/health |
{ status: "ok", timestamp } |
agents-platform/
├── docker-compose.yml # App + Qdrant + Ollama
├── Dockerfile # Multi-stage: build React, serve with Express
├── .env.example
│
├── server/
│ ├── index.js # Express app, middleware, route wiring
│ ├── db.js # SQLite schema init, migration
│ ├── seed.js # 20 agent persona definitions
│ ├── demo.js # Demo mode seed data
│ ├── executor.js # SSH dispatch, parallel/sequential/meeting modes
│ ├── scheduler.js # node-cron scheduler, Discord notifications
│ ├── agency-sync.js # Agent sync utilities
│ ├── mcp-registry.js # MCP server definitions
│ ├── safety-prompt.js # Prepended safety context for SSH runs
│ ├── rag/
│ │ ├── qdrant.js # Qdrant client wrapper
│ │ ├── embeddings.js # Ollama embedding client
│ │ ├── splitter.js # Recursive text chunker
│ │ ├── ingest.js # Document ingestion pipeline
│ │ ├── chat.js # RAG search and chat
│ │ └── loaders/ # Pluggable loaders (md, yaml, tf, url, transcript)
│ ├── workflows/
│ │ ├── state.js # LangGraph state schema
│ │ ├── tools.js # LangChain tools (kubectl, file read, RAG)
│ │ ├── router.js # Task classifier (Haiku)
│ │ ├── graphs.js # LangGraph graph definitions
│ │ └── runner.js # Graph execution engine
│ ├── eval/
│ │ └── runner.js # Eval suite runner with LLM judge
│ ├── observability/
│ │ └── telemetry.js # Cost calculator, trace middleware
│ └── routes/
│ ├── agents.js # Agent CRUD
│ ├── agency.js # Agency operations
│ ├── schedules.js # Cron schedule management
│ ├── runs.js # Run history
│ ├── apps.js # App discovery
│ ├── rag.js # RAG ingest and query
│ ├── workflows.js # LangGraph workflow execution
│ ├── observability.js # Telemetry and dashboard
│ └── eval.js # Evaluation suites and LLM judge
│
├── src/
│ ├── App.jsx # React Router setup, lazy page loading
│ ├── index.css # Tailwind + glassmorphism custom styles
│ ├── components/
│ │ ├── Layout.jsx # Nav header with 7 sections
│ │ ├── AgentCard.jsx
│ │ ├── AgentAvatar.jsx # 20 unique inline SVG avatars
│ │ ├── rag/ # IngestPanel, SearchPanel, ChatPanel, SourceList
│ │ └── workflows/ # GraphView (SVG route visualization)
│ └── pages/
│ ├── Home.jsx # Agent grid, search, category filter
│ ├── AgentProfile.jsx # Full agent detail, activate modal
│ ├── AgencyAgentProfile.jsx
│ ├── ComposePage.jsx # Multi-agent composition
│ ├── SchedulesPage.jsx
│ ├── ScheduleDetailPage.jsx
│ ├── AllRunsPage.jsx
│ ├── RunDetailPage.jsx
│ ├── RagPlayground.jsx # RAG document loader + query UI
│ ├── WorkflowsPage.jsx
│ ├── ObservabilityPage.jsx
│ └── EvalPage.jsx
│
└── test/
└── rag-smoke.js # RAG integration smoke test
# Install dependencies
npm install
# Start backend (watches for changes)
npm run dev:server
# Start frontend dev server (in a second terminal)
npm run dev
# Build for production
npm run build
npm startThe Vite dev server proxies /api to :3001, so both servers run simultaneously without CORS issues.
Local Qdrant and Ollama:
docker run -p 6333:6333 qdrant/qdrant
docker run -p 11434:11434 ollama/ollama
docker exec <ollama-container> ollama pull nomic-embed-textSet QDRANT_URL and OLLAMA_URL in your .env.
SSH Dispatch:
Set SSH_TARGET=user@your-host and ENABLE_SCHEDULER=true in .env. The target host must have Claude Code installed and accessible via SSH key auth. Set SSH_KEY_PATH if the key is not at the default location.
MIT