ContextAgora.com
The context layer for AI agents. Stop re-explaining your product to AI.
Role: Founder & Lead Developer
Timeline: ongoing
Technologies: Python (FastAPI), React (Vite, TanStack), SQLite, Docker, Claude Code, Varlock
Project Overview
ContextAgora is a self-hostable system that gives AI coding agents the operational context of your product: API docs, runbooks, integration keys, live data, so they can act, not just talk. Modules live in your GitHub repo; the agent loads only what each task needs.
The Challenge
Generic AI agents arrive at every conversation amnesic about your product. Teams spend hours re-explaining schemas, runbooks, and integration quirks before the agent can do anything useful, and even then exports go stale and credentials leak across tasks.
AI Integration Highlights
- Module-based context loading: select what the agent sees per task instead of dumping the whole codebase into the window
- Per-task secret loading via Varlock; keys are loaded only while a task runs and discarded afterward, so the agent has no standing access between runs
- Live integrations rather than stale exports: the agent queries real APIs (databases, ticketing systems, billing) on every request
- Self-improving support workflows: each resolution becomes a reusable module the next agent reads first
Key Features
- Context modules from GitHub: modules, prompts, and runbooks live in your repo and load on demand.
- Integration registry: add a new integration in ~5 minutes by dropping a module into the workspace.
- Workflows & jobs: scheduled cron jobs and multi-step workflows run as background tasks.
- Slash commands & mentions: drive the agent from chat with
/commandsand@filementions. - Self-hosted or SaaS: runs as a Docker container on your own infrastructure, or fully hosted.
Results
In production with MAAT (SaaS for martial arts gyms):
- 90% faster support resolution
- One agent fronting all internal integrations
- Live data on every query, no stale exports