<2-3 sentences summarising what the project addresses and solves, the target audience, and the benefit.> Aegis is an Architecture-Driven Development (ADD) method pack for AI coding agents, enhancing reliability for long tasks. It enforces architecture-awareness, evidence-verified development, and drift-checking, targeting AI hosts and users to improve software development discipline and address process failures.

How It Works

Aegis mandates agents understand project baselines and architecture boundaries before changes. It integrates evidence-driven governance, TLREF execution flow, and dual-track repair/retirement rules. This ensures agents frame tasks with impact awareness, maintain evidence, and manage long work via checkpoints and drift checks, providing stricter discipline without a new runtime.

Quick Start & Requirements

Integration involves adding Aegis to compatible AI coding hosts (Codex, OpenCode, Claude Code) via native plugin/skill paths. Setup varies by host, often using git clone or plugin configurations. Post-installation and host restart, verification uses python scripts/aegis-doctor.py --write-config --json, requiring "ok": true, "workspaceSupport": "available", and "configStatus": "configured". Host guides and compatibility matrices are available.

Highlighted Details

• Architecture-Driven Development (ADD): Prioritizes project context and verification paths before changes.
• Evidence-Verified Workflows: Requires fresh verification evidence before completion claims.
• Long-Task Continuity: Supports resumability and drift checks via checkpoints.
• Dual-Track Governance: Separate tracks for bug fixes (repair) and feature changes (retirement).
• Multi-Host Compatibility: Designed for integration with various AI coding agents.

Maintenance & Community

Builds on "Superpowers" methodology, acknowledging Jesse Vincent and Matt Pocock. Specific community channels or active maintainer details are not provided in the README excerpt.

Licensing & Compatibility

Released under the MIT License, permitting broad commercial and closed-source integration. Core design goal is multi-host compatibility, with status updates for AI coding agents documented.

Limitations & Caveats

Aegis is a method pack, not an authoritative runtime core or decision system; its outputs are advisory. It lacks host watchdogs or automatic retries. Effectiveness relies on a clear project baseline; performance may be less stable without one. Host compatibility varies, with some platforms pending verification.