ouroboros_start_auto
Start or resume an Ouroboros automation in the background, obtaining session and job IDs to monitor progress and retrieve final results.
Instructions
Start ooo auto in the background and return auto_session_id + job_id immediately. Resume with the returned auto_session_id; poll with ouroboros_job_status / ouroboros_job_wait and read final state via ouroboros_job_result.
Input Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| cwd | No | Working directory | |
| goal | No | Goal/task for ooo auto | |
| domain | No | Optional domain profile name (for example, coding) | |
| resume | No | Auto session id to resume | |
| skip_run | No | Stop after A-grade Seed | |
| attach_job | No | Attach an externally verified job id to an unknown run handoff | |
| attach_source | No | Source label for an attached run handle | |
| commit_policy | No | Checkpoint commit policy: ac_checkpoint, final_only, or none | |
| reconcile_run | No | Try to reconcile an unknown run handoff without starting a duplicate run | |
| attach_session | No | Attach an externally verified run session id to an unknown run handoff | |
| efficiency_mode | No | Execution efficiency policy: adaptive may use lower-cost child tiers with recovery escalation; quality_first keeps child ACs at the parent starting tier. Default: adaptive. | |
| worktree_policy | No | Worktree isolation policy: auto, always, current, or none | |
| attach_execution | No | Attach an externally verified execution id to an unknown run handoff | |
| complete_product | No | When true, chain RUN → RALPH_HANDOFF after a successful run handoff so a single ouroboros_auto invocation iterates Ralph until QA passes, convergence, or a budget bound trips. Defaults to false (opt-in). | |
| reconcile_source | No | Source label for run handoff reconciliation | |
| user_preferences | No | Caller-supplied user preferences keyed by ledger section name (e.g. runtime_context, constraints, non_goals). The Driver tags matching answers with [from-auto][user_preference] in the ledger. Keys must be valid ledger section names; values must be non-empty strings or non-empty lists of strings/numbers. On resume, null/empty values clear the persisted preference for that section. | |
| max_repair_rounds | No | Max repair rounds | |
| frugality_assurance | No | Frugality assurance: off, observe, or explicitly authorized strict. Defaults from efficiency_mode and is immutable on resume. | |
| max_interview_rounds | No | Max interview rounds | |
| pipeline_timeout_seconds | No | Top-level pipeline deadline in seconds. Defaults to 7200s for new sessions. Range: 60-86400. Cannot be changed on resume; the deadline is preserved across process restarts. |