Activity and Job YAML

Activity Envelope

schemaVersion: 2
kind: Activity
metadata:
  name: example_activity
spec:
  type: deterministic
  description: Run a registered deterministic action.
  action: example_action
  input_schema_json:
    type: object
    properties: {}
  output_schema_json:
    type: object
    properties:
      status:
        type: string

Activity Types

Type	Required fields	Notes
`agent_loop`	`instruction`; optional `tools`, `provider`, `model`, `wall_clock_timeout_seconds`	`tools` is the activity baseline. For task-backed dispatch Orbit adds exact `task.required_tools`, deduplicates the union, and rejects invalid requirements before provider launch. Agent execution uses the CLI path only. `backend: cli` still parses and is ignored; `http` and `auto` fail catalog load. `orbit doctor --fix-retired-activity-backends` removes those retired keys. `max_iterations` is inert.
`deterministic`	`action`; optional `config`	Runs a registered deterministic action.

The computed effective list is serialized as tools in the CLI execution envelope and exported as ORBIT_ACTIVITY_TOOLS. The task-requested list is serialized separately as required_tools; it is immutable after task creation, and audit evidence records both lists. Allowlist inclusion never substitutes for runtime role, capability, policy, filesystem, subprocess, or authentication checks.

Job Envelope

schemaVersion: 2
kind: Job
metadata:
  name: example_job
spec:
  state: enabled
  max_active_runs: 1
  kind: workflow
  steps:
    - id: run_action
      target: activity:deterministic_reference

Step Bodies

Reference an activity:

- id: assess
  target: activity:agent_assess_diff

Inline a full activity spec:

- id: run_action
  spec:
    type: deterministic
    action: example_action
    config: {}

Run branches in parallel:

- id: parallel_assessment
  parallel:
    join: { mode: all }
    branches:
      - id: branch_a
        target: activity:assess_a
      - id: branch_b
        target: activity:assess_b

Modifiers

Each step may include when and retry.

retry:
  max_attempts: 3
  initial_backoff_ms: 500
  backoff_cap_ms: 5000
  backoff_strategy: exponential