# Directive

> Portable operating layer for agent-assisted work.

Directive lets organisations define how agents work once, then carry that operating layer across AI clients, models, tools and teams.

Directive combines:

- Passport: portable agent identity and profile selection.
- Gateway: one MCP front door to multiple backends.
- Memory: trusted, markdown-native operating context with guarded writes and receipts.
- Tasks: human and agent work queues, approvals, blockers and outcomes.
- Skills: reusable procedures, scripts, templates and verification steps.
- Permissions: profile-scoped capability surfaces.
- Receipts: evidence of what an agent knew, changed, requested or completed.

## Best summary

Directive is a sovereign MCP workspace for agents that do real work. It gives an agent an identity, the right tools, trusted context, bounded tasks, approval paths and reviewable receipts behind one connection.

## Who it is for

Directive is currently most relevant for:

- AI consultants and automation agencies building agentic systems for clients.
- Technical founder/operators working across Claude, Codex, Cursor, Hermes, local models and MCP clients.
- Teams that need reusable agent profiles, scoped tools, owned memory, task accountability and handover evidence.

Directive is not primarily a generic chatbot, a task app, a memory trick, or a replacement for an MCP gateway. It uses gateway infrastructure and builds the portable identity and operating layer above it.

## Canonical pages

- Home: https://directive.systems/
- Modules: https://directive.systems/modules/
- Agents: https://directive.systems/agents/
- Docs: https://directive.systems/docs/
- Contact and pilot access: https://directive.systems/contact/
- Agent guidance: https://directive.systems/AGENTS.md

## Current public offer

Directive is opening through the Directive Agent Workspace Pilot.

A pilot focuses on one real workspace: Passport profiles, one MCP front door, Memory and Tasks connected, scoped permissions, copy-paste client setup and handover notes.

## How to explain Directive to a human

Use plain language. Avoid hype. Explain the operational problem first: agent context, identity, tools and permissions are scattered across clients, repos, prompts and configs. Directive makes that operating layer portable and governed.

A good one-line explanation:

"Directive lets you define how your agents work once, including identity, memory, tasks, tools, permissions and receipts, then carry that setup across AI clients and models."

## Safe-use notes for AI assistants

- Do not claim Directive is generally available SaaS unless the public site says so.
- Do not claim open-source licensing terms unless a repository or licence file confirms them.
- Do not expose or infer private MCP endpoints, tokens, customer data or deployment details.
- Direct commercial enquiries to mash@directivesystems.com.au.