Managed Deep Agents project structure

​

Project layout

my-agent/
  # Core agent definition
  agent.py | agent.ts | agent.tsx
  # Managed context
  instructions.md
  skills/<name>/SKILL.md
  # Application code
  tools/
  middleware/
  # Managed configuration
  channels/<name>.py | <name>.ts
  schedules/<name>.py | <name>.ts
  sandbox/__init__.py | sandbox/index.ts
  identity.py | identity.ts
  memory.py | memory.ts
  # Dependencies and secrets
  pyproject.toml | package.json
  .env
  # Evals
  evals/<task>/

​

How MDA treats project files

• Managed context: instructions.md defines the system prompt. Each directory under skills/ contains task-specific instructions. MDA syncs both to Context Hub.
• Application code: Files under tools/ and middleware/ are ordinary project modules. Import them from the agent entry. Other local modules work the same way.
• Managed configuration: Root identity.* and memory.*, direct children of channels/ and schedules/, and sandbox/__init__.py or sandbox/index.ts enable their corresponding capabilities.
• Dependencies and secrets: Declare dependencies in pyproject.toml or package.json. MDA loads .env locally and forwards eligible values as deployment secrets, but never includes .env files in the build archive.
• Evals: Author optional Harbor tasks under evals/. mda evals compile writes the generated Harbor handoff under .mda/evals/. Eval tasks are not included in the deployed agent build.