Campaign Builder home

Bruno execution recovery

How Bruno records partial SFMC execution.

Durable Bruno runs make partial draft production reviewable: stage state, journals, artifacts, rollback metadata, resume constraints, and manual Marketing Cloud follow-up stay visible after execution succeeds, fails, or partially rolls back.

Recovery model

Durable execution leaves evidence when work partially completes.

Bruno execution is designed for technical review after both dry runs and confirmed runs. The implementation records stage state, journals, artifacts, rollback metadata, and resume constraints so evaluators can see what happened before continuing in Marketing Cloud.

Run journal

Each durable campaign execution run stores request-aware journal entries such as run creation, resume events, stage transitions, warnings, and failure details for later inspection.

Artifact ledger

Created and reused outputs are tracked as run artifacts with stage, ID, customer key, source, and rollback eligibility metadata where the underlying SFMC operation returns enough detail.

Rollback scope

Rollback is limited to created artifacts with matching run provenance. It can attempt supported cleanup for Data Extensions, Content Builder assets, email assets, preview drafts, and folders, then returns warnings or failures instead of hiding partial cleanup.

Resume guard

resumeRunId only continues a run when the requested mode and campaignSpec match the stored run, which helps avoid duplicating SFMC work after a transient execution failure.

Evaluator review path

What to inspect before deciding the next action.

Partial execution review should begin with the run record, not with assumptions about what Marketing Cloud changed. The useful question is which artifacts were proposed, created, reused, cleaned up, or left for manual review.

Start with the stored run

Review the mode, campaignSpec, plan stages, stageState, summary, journal, artifacts, and rollback metadata before deciding whether to resume, retry outside Campaign Builder, or continue manually in Marketing Cloud.

Separate created and reused artifacts

The artifact ledger distinguishes outputs Campaign Builder created from assets it reused. Rollback eligibility is intentionally narrower than the full artifact list.

Read rollback results as evidence

Supported cleanup can delete matching run-created Data Extensions, Content Builder assets, email assets, preview draft data, and folders when identifiers and endpoints are available.

Keep final launch decisions in SFMC

Recovery evidence helps teams understand draft production state; it does not activate journeys, approve sends, or replace Salesforce Marketing Cloud review.

Route-backed checks

What the implementation makes reviewable.

The recovery model is grounded in the Bruno run store, stage result normalization, rollback helper, and campaign orchestrator rather than a generic AI chat transcript.

Durable run records keep stage state, journal entries, artifact ledgers, rollback metadata, status summaries, request IDs, and account-scoped context.
Stage results normalize completed, reused, skipped, and failed outcomes so partial progress can be reviewed after Factory-backed folders, Data Extensions, email, preview, and validation stages.
Rollback attempts only use artifacts marked created, rollbackEligible, and tied to the same runId, then report deleted items, warnings, failures, and partial rollback status.
Resume uses resumeRunId only when the stored execution mode and campaignSpec match the current request, which helps avoid duplicate SFMC draft production.

Technical evaluator FAQ

The recovery boundaries to verify.

Recovery should reduce uncertainty after partial execution. It should not be positioned as certain cleanup, automatic send approval, or a replacement for SFMC inspection.

Does rollback guarantee a clean SFMC reset?

No. Campaign Builder records and attempts supported cleanup for eligible created artifacts, then exposes warnings or failures when an identifier, endpoint, provenance match, or helper is unavailable.

What should an SFMC admin inspect after a failed run?

Inspect the journal, stageState, artifacts, rollback warnings, requestId, created versus reused outputs, and any draft assets or Data Extensions that now exist in Marketing Cloud.

When should a team resume instead of restarting?

Resume is appropriate only when the stored run matches the requested mode and campaignSpec. Otherwise the safer path is to review the artifacts and run a new dry run before any confirmed execution.

Evaluation takeaway

Partial execution should leave a trace an SFMC admin can inspect.

Campaign Builder's Bruno recovery model supports safer evaluation because every confirmed execution can be reviewed through run state, stage results, artifacts, rollback status, and the continued manual launch boundary in Marketing Cloud.

Review Bruno orchestrationCheck SFMC write boundaries