AGENTS.md Standard

The AGENTS.md file provides a vendor-neutral mechanism for configuring AI behavioral logic across diverse coding platforms. It enables you to codify project-specific standards and tactical instructions that Sypha - and other AI agents - should intuitively follow.

The Purpose of AGENTS.md

AGENTS.md functions as a "README for AI," serving as an open standard for project-level agent configuration. By placing this Markdown asset at your project root, you provide AI participants with the technical context and procedural guidelines necessary for your specific environment.

Strategic Advantages

• Cross-Platform Portability: Recognized across multiple AI-driven engineering environments.
• In-Repo Governance: Persists within your version control system alongside your primary source code.
• Team-Wide Sync: Ensures that every contributor's AI assistant adheres to the same architectural and stylistic constraints.

Asset Placement

• Repository Root: Place the primary AGENTS.md file at the root of your project.
• Contextual Subdirectories: You can deploy secondary AGENTS.md files within specific modules (e.g., api/v2/AGENTS.md) to define localized logic subsets.

[!WARNING] The filename is case-sensitive and must be entirely uppercase: AGENTS.md.

Integrity Protection

To prevent unauthorized logic modifications, AGENTS.md is a write-protected asset within the Sypha ecosystem. The agent cannot alter its content without your explicit, manual authorization, ensuring your project's AI configuration remains secure.

Recommended Architectural Structure

# Project Architectural Guidelines

## Syntax & Style
- Enforce strict typing in all TypeScript assets.
- Utilize functional programming patterns for data transformations.

## System Architecture
- Adhere to the hexagonal architecture pattern.
- Decouple domain logic from infrastructure adapters.

## Quality Assurance
- Implement comprehensive integration tests for all new endpoints.

Logic Loading Hierarchy

Sypha parses AGENTS.md at runtime and integrates its directives into the active reasoning context. It is prioritized within the following rule stack:

1. Persona-Specific Rules (.sypha/rules-{mode}/) [Highest Priority]
2. Project Rules (.sypha/rules/)
3. AGENTS.md Directives
4. Global Rules (~/.sypha/rules/) [Baseline Priority]

Control Toggles

Support for the AGENTS.md standard is active by default. To disable this feature, update your settings.json:

{
  "sypha.useAgentRules": false
}