AGENTINK Developer Portal
Production Environment Active

AgentInk: The agent that remembers, adapts, and overcomes.

A stateful, self-healing, and proactive developer agent. AgentInk orchestrates specialized child agents, automates complex workflows, and integrates deeply with your development stack.

Core Platform Usage

AgentInk is structured around highly functional modules designed to streamline development, infrastructure, and spatial analysis.

Live Telemetry & Android Automation

Monitor autonomous tasks executed on Android emulators in real-time. View live screenshots, step-by-step execution logs, and stream telemetry data directly to the Control Center.

Module: Telemetry Open Telemetry

Meeting Facilitator

Schedule, launch, and facilitate Google Meet, Zoom, or Teams meetings. AgentInk joins the call, walks the agenda, time-boxes items, and posts evidence-quoted minutes directly to Slack.

Module: Meetings Open Facilitator

Spatial Siting Explorer

Execute complex spatial siting searches using the Moonshot API. Find optimal locations based on custom criteria (e.g., proximity to substations, zoning, floodplains) and compile results into Google Earth KMZ layers.

Module: Siting Open Explorer

Parent-Child Agent Orchestration

Spawn, oversee, and manage specialized child agents (like infabot in #infrastructure) directly from Slack. Child agents operate in secure sandboxes with dedicated system instructions.

Module: Orchestration Manage Agents

Interactive GIS Siting Map

Real-time visualization of Bastrop County substations, 800m proximity buffers, and candidate parcels compiled from the Moonshot API.

Substations (⚡)
800m Proximity Buffer
Candidate Parcels

Development Workflow & Order of Operations

To maintain platform stability and security, all developers and agents must strictly adhere to the following workflow.

STEP 01

Proposal & Discussion

Before writing any code, propose the change in the appropriate Slack channel or open a GitHub issue to align on architecture and scope.

STEP 02

Branch Creation

Create a dedicated feature or bugfix branch from the latest main branch. Direct commits to main are strictly prohibited.

STEP 03

Implementation & Testing

Implement changes following clean, modular, and documented standards. Run all unit and smoke tests locally to ensure no regressions.

STEP 04

Pull Request (PR)

Push your branch to GitHub and open a Pull Request (PR) against the main branch with a clear description and test results.

STEP 05

Approved Review

Request a review from an approved developer (e.g., Gregory Reeves or Rob). All PRs require at least one approved review before merging.

STEP 06

Merge & Deploy

Merges to main are performed only by authorized personnel. Deploy the changes and run post-deployment smoke tests.

System Architecture & Tech Stack

AgentInk is built on a robust, enterprise-grade architecture designed for high availability, security, and continuous self-evolution.

Core Infrastructure

  • GCP VM: Hosted on a secure Google Cloud Platform virtual machine.
  • Caddy Web Server: Handles reverse-proxying and auto-provisions SSL certificates.
  • FastAPI Backend: High-performance Python web framework serving the Control Center API.

Stateful Memory

  • Neo4j Knowledge Graph: Stores system state, task dependencies, and child agent logs.
  • PostgreSQL: Secure database with Row-Level Security (RLS) for encrypted user secrets and mirror profiles.
  • Persistent Reflections: Tracks developer preferences, active tasks, and milestones.

Stack Integrations

  • Slack Socket Mode: Real-time, secure communication bridge for parent and child agents.
  • Linear & Google Drive: Deep integration with project management and document storage.
  • Google OAuth: Secure, domain-restricted authentication for the Control Center.

Persistent User Reflections

AgentInk doesn't just execute commands; it remembers. By passively listening to updates and analyzing interactions, AgentInk builds a persistent reflection profile for each developer.

This profile tracks your coding style, active tasks, milestones, and preferences, allowing AgentInk to tailor its assistance and proactively support your engineering workflow.

gregory.reeves_mirror.md Markdown

# Reflection Profile: gregory.reeves

## Preferences

- Language: Python

- Coding Style: Clean, modular, thoroughly documented.

## Active Tasks

- Implementing Phase 1.5 (Android Emulator Agent & Live Telemetry Dashboard)

## Tagline

"AgentInk: The agent that remembers, adapts, and overcomes."