Skip to content

Latest commit

 

History

History
18 lines (14 loc) · 1.14 KB

File metadata and controls

18 lines (14 loc) · 1.14 KB

AGENTS.md

Operational Protocols

1. Resource Verification

Before attempting to read any synaptic data (registries, topologies), agents MUST verify the existence and integrity of the resources/ directory. Use dynamic path resolution relative to the plugin root.

2. Neural Processing Standard (Guard → Do → Verify)

Start every file with a Why/What/How docblock; add only timeless structural comments that explain architecture or invariants, never TODOs or status notes.
Keep modules focused, under ~500 lines, and composed of small, single-responsibility pieces; favor composition over inheritance and avoid circular dependencies.
Define contracts before coding: spell out inputs, outputs, error modes, performance budgets, side effects, and telemetry in comments and tests.
Follow Guard → Do → Verify flow: validate and normalize inputs, perform core work cleanly, then assert outputs and invariants; freeze or clone boundary data.

Core Concepts

  • Core: A specialized AI workflow triggered by the First Post in a designated subcategory.
  • Core Seed: The Output generated by the Core workflow (the First Reply).