Reliant

Reliant is an AI coding assistant that goes beyond chat. While you can interact with your codebase through natural conversation, Reliant’s real power comes from programmable workflows that let you automate complex, multi-step development tasks—and check them into git alongside your code.

What Makes Reliant Different

Most AI coding tools give you a chat interface. You ask questions, get answers, and manually orchestrate each step of a complex task. Reliant starts there but doesn’t stop.

Workflows let you define repeatable patterns that combine AI reasoning with deterministic logic. Need to review a PR, run tests, analyze failures, and create fix suggestions? That’s not a conversation—it’s a workflow. You define it once, trigger it when needed, and let agents handle the execution while you focus on decisions that matter.

This isn’t about replacing chat. The conversational interface remains the primary way you interact with Reliant. Workflows extend that interaction into territory that chat alone can’t reach: tasks that span multiple files, require conditional logic, or benefit from parallel execution.

Key Capabilities

Chat with Your Codebase

Reliant understands your project structure, reads your files, and maintains context across your conversation. Ask questions about how components connect, request explanations of complex logic, or get help debugging issues. The AI has access to tools for searching, reading, and modifying your code—turning conversation into action.

Programmable Workflows

Workflows are YAML configurations that define multi-step processes. Each step can invoke AI agents, run shell commands, make HTTP requests, or execute conditional logic. You can loop over collections, branch based on results, and compose complex behaviors from simple building blocks.

Unlike chat, workflows are reproducible. They execute the same way every time, making them suitable for CI/CD integration, team-wide standards, and tasks where consistency matters more than flexibility.

Multi-Agent Patterns

Some problems benefit from multiple perspectives. Reliant supports agent patterns like debate, where agents argue different positions before converging on a solution; parallel execution, where multiple agents work simultaneously on independent subtasks; and conditional loops, where agents iterate until a condition is met.

These patterns aren’t theoretical—they’re configurations you define and trigger. A code review workflow might spawn three agents to evaluate different aspects of a change, then synthesize their findings into a unified report.

Config as Code

Workflows live in your repository as .reliant/workflows/*.yaml files. They’re versioned with your code, reviewed in PRs, and shared across your team. Project-specific conventions become encoded in workflows rather than tribal knowledge.

Custom presets let you define agent personalities and tool configurations that match your project’s needs. A security-focused preset might include specific tools and prompting for vulnerability analysis; a documentation preset might emphasize clarity and completeness.

Quick Links

Getting Started

Workflows

Reference

Configuration

  • Presets — Reusable parameter bundles
  • MCP Servers — Extend Reliant with external tools
  • Settings — API keys, providers, and preferences

Community

Installation