Agena

data-ai MCP Server

Agentic AI platform with pixel agent technology — autonomous code generation, PR automation & multi-tenant SaaS

Verified
data-aidata-ai
6 views79 stars16 forksMIT

Why This Matters

Discovered via agent-topic:crewai and last synced 3mo ago.

Verified
Source
agent-topic:crewai
Stars
79
Last synced
3mo ago
Install
Check source

Install

Install instructions not detected yet

Check the source repository for the latest setup steps.

View source instructions
41
Tools
0
Resources
0
Prompts
Standard I/O
Transport

Available Tools (41)

GITHUB_OWNER

GitHub org/user

LLM_LARGE_MODEL

Model for complex tasks

frontend_blue

ai_agent_frontend_blue

Reviewer

Large + reasoning

Variable

Description

REDIS_URL

Redis connection URL

GITHUB_TOKEN

GitHub PAT for PR creation

LLM_MODEL

Default LLM model

frontend

ai_agent_frontend

Developer

Large (128K output)

Service

Container

OPENAI_API_KEY

OpenAI API key — global default if set

QDRANT_ENABLED

Enable vector memory

worker

ai_agent_worker

Method

Path

GITHUB_REPO

Default repo name

LLM_SMALL_MODEL

Model for simple tasks

JWT_SECRET_KEY

JWT signing secret

QDRANT_URL

Qdrant server URL

cli-bridge

(host process)

MYSQL_DATABASE

Database name

redis

ai_agent_redis

Agent

Model

Challenge

How AGENA Solves It

Layer

Technology

mysql

ai_agent_mysql

stop

status

Unlimited

Custom models, SSO, dedicated support

frontend_green

ai_agent_frontend_green

MYSQL_HOST

MySQL host

qdrant

ai_agent_qdrant

Finalizer

Small

backend

ai_agent_api

PUT

`/preferences/prompts`

Plan

Tasks/Month

queued

...] agena task show <id> agena task logs <id> agena task create -t "Fix login bug" --assign # Team skill catalog agena skill list [-q <query>] [-t <pattern_type>] agena skill show <id> agena skill search "nullable pointer panic" # vector search agena skill delete <id> -y # Sprint refinement (Qdrant-grounded SP estimates) agena refinement backfill -p MyProject -t MyTeam --days 730 agena refinement backfill-status agena refinement history [--sp 5] [-q auth] agena refinement analyze -p MyProject -t MyTeam --sprint-path '...' ``` - **Device-code OAuth** — no JWT copy-paste; browser opens, you confirm a 6-digit code. - **Keychain credential storage** — JWT held in macOS Keychain / libsecret / Windows Credential Manager when available. - **Bundled bridge** — `agena daemon start` ships with its own copy of `bridge-server.mjs`, so `npm install -g @agenaai/cli` is self-contained. - **Homebrew tap** — `aozyildirim/tap/agena` auto-updates via GoReleaser on every CLI release. See [`packages/cli/README.md`](packages/cli/README.md) for the full command reference. --- ## Quick Start ### 1. Clone and configure ```bash git clone https://github.com/aozyildirim/Agena.git cd Agena cp .env.example .env cp frontend/.env.example frontend/.env.local ``` ### 2. Set required environment variables in `.env` The only hard requirement is the JWT signing secret — everything else (LLM provider, GitHub / Azure DevOps / Jira credentials) is configured per-organization through the dashboard at `/dashboard/integrations` once the stack is running, so the same deployment can serve teams that use OpenAI alongside teams that use Claude CLI. ```env JWT_SECRET_KEY=your-secret-key # Optional — only set if you want a global default LLM key. Most users # leave these empty and configure providers per-org from the UI. # OPENAI_API_KEY=sk-... # GEMINI_API_KEY=... ``` ### 3. Start all services ```bash ./start.sh ``` This starts Docker services (backend, worker, frontend, MySQL, Redis, Qdrant) and the **CLI bridge on the host** for Claude/Codex CLI authentication via system keychain. > **Note:** Use `start.sh` instead of `docker-compose up`. The CLI bridge must run on the host (not in Docker) to access Claude/Codex CLI auth. `./stop.sh` stops everything. ### 4. Access

GET

`/health`

GEMINI_API_KEY

Google Gemini API key — global default if set

MAX_WORKERS

Concurrent worker tasks

POST

`/billing/stripe/checkout`

5

Basic pipeline, 1 integration