How should teams manage shared configuration across AI coding agents?

Question

I have been thinking about a problem that sits between developer tooling and AI infrastructure.

Suppose a repository has shared instructions, rules, or manifests. Different agent environments may need different native representations of that same source.

Copying files manually works until:

• one target gets updated and another does not;
• the team cannot tell which environment is active;
• generated files drift from the intended spec;
• adding another adapter requires more custom logic.

I built agent-compat to experiment with a different model:

---

Installation & Setup

npm install @jstn-sdk/agents

How It Works

1. Detect

Automatically discovers which agent environments are present in a project:

const detected = await Agents.detect("./my-project");
// → [{ id: "cursor", confidence: 0.95 }, { id: "codex-cli", confidence: 0.92 }]

2. Compile

Transforms a canonical manifest into native files for each environment:

const manifest = {
  version: 1,
  project: { name: "my-app", stack: ["typescript"] },
  instructions: ["Run tests before completion"],
  skills: { "code-review": { description: "Review PRs" } }
};

const result = await Agents.compile(manifest, {
  targets: ["cursor", "codex-cli", "pi"],
  output: "./my-project"
});
// → { files: [".cursor/rules/agents.mdc", "AGENTS.md", ".pi/skills/review/SKILL.md"] }

3. Validate

Checks generated files against official specifications:

const report = await Agents.validate("./my-project");
// → { cursor: "✓", "codex-cli": "✓", pi: "◐", summary: { ... } }

---

Architecture

agents/
├── detect()     ← Environment detection
├── compile()    ← Manifest → native files
├── validate()   ← Spec compliance check
└── registry     ← Plugin adapter system

It is a library-first SDK with a companion CLI, so the workflow can be embedded into CI, internal tooling, or editor integrations.

The question I am exploring is not only “how do we support more agents?”

It is:

Should agent configuration be treated like source code that needs
compilation and validation?

Repository: https://github.com/JustineDevs/agent-compat

I would especially like feedback from people maintaining monorepos, shared developer instructions, or multiple coding-agent workflows.

• Do you keep one source of truth for agent instructions?

• Where does validation happen today?

• Would you prefer generated native files or runtime resolution?

• What would make an adapter ecosystem useful to you?