Skip to content
AssureSwarm Docs

Agent Operating Guide

/coach-workflow-export

Download a workflow's complete data, every step result, form value, and attached document, into a local directory. Read-only.

Bundle a workflow instance’s complete data into a local directory: every step’s saved result, every form’s submitted values, and every uploaded or linked document. The skill walks the workflow, downloads each document, and writes a structured package with a manifest you can archive or share. It’s read-only: it never goes through suggest_change and never changes the workflow.

When to use

Section titled “When to use”

When you want a raw local snapshot of a workflow: to archive, hand off, or inspect offline. For a redaction-gated formal export across multiple items, use /coach-export-package instead (it pipes through the redactor); this is the simpler raw-dump variant.

Inputs

Section titled “Inputs”
Flag Required Notes
--workflow-id <id> or --workflow-name <text> No The workflow to export. If absent, inferred from your current view.
--output-dir <path> No Where to write the package. Defaults to a timestamped folder under .coworkcanvas/exports/.

Example

Section titled “Example”

/coach-workflow-export --workflow-name "Q2 Fieldwork" resolves the workflow and writes a package: workflow.json (metadata and steps), a steps/<id>.json per step (result body, form values and fields, document metadata), the downloaded documents under documents/<step-id>/, and a top-level MANIFEST.json. The summary reads something like “Workflow exported: 7 steps, 14 documents, 3.2 MB.”

Good to know

Section titled “Good to know”
  • No redaction. The output contains every field exactly as stored: internal-only fields, review notes, and coaching comments included. If it’s leaving the team, use /coach-export-package, which pipes through the redactor.
  • Binary documents are base64-encoded. Text documents (markdown, csv, txt) write as-is; binaries (pdf, docx, xlsx, images) write as .b64 with the mime type noted in a per-step index. Decode them with base64 -d.
  • Linked documents capture the URL, not the bytes. For documents linked as external URLs, the export records the URL, filename, and type: it doesn’t fetch the external content (that would need credentials the platform doesn’t have).
  • Safe by default. The output directory is created if missing, but the export fails fast rather than overwriting a non-empty one, and it surfaces the estimated size: a workflow with many PDFs can reach hundreds of MB.
Section titled “Related”