[01] LOOP
OperationsintervalHardenedTemplate

AI Document Workflows for Enterprises on Autopilot

Inspired by EigenPal — AI Document Workflows for Enterprises. Loop until the workflow is current, exceptions are owned, and human sign-off is captured where required.

Inspired by EigenPal

by Trooper

Kickoff prompt
/loop 30m Start the "AI Document Workflows for Enterprises on Autopilot" loop.

Inspired by EigenPal (https://www.eigenpal.com).

Goal: open work triaged, exceptions owned, and core operations workflow current with audit trail
Max iterations: 20
Between iterations run: Report open queue items, stale tasks, failed automations, and items awaiting human approval for EigenPal
Exit when: zero open items without owner or explicit escalation, all external actions approved or sent, and systems of record current

Step 1 — Intake documents: Collect uploads from portal, email, or integrations; track cases.
Step 2 — Extract structured data: AI reads documents with per-field confidence scores.
Step 3 — Build master record: Merge fields; resolve conflicts across source documents.
Step 4 — Human review: Route low-confidence or judgment calls for sign-off.
Step 5 — Deliver outputs: Fill templates, export packages, or push to downstream systems.

## Before you start

Connect plugins:
- GitHub (required) — Read branches, PRs, reviews, checks, workflow runs, and source diffs.
- Gmail (required) — Read incoming mail and prepare safe draft replies for approval.
- Slack (required) — Post summaries, approvals, blockers, and handoff updates.

Attach skills:
- Loop runner (required) — Self-pace iterations, run the check between passes, and stop only on the exit condition.
- Code change + local verification (optional) — Edit code safely, run commands, and keep changes scoped.
- CI debugging (optional) — Read failing checks, logs, and the smallest actionable root cause.
- Approval workflows (optional) — Keep outbound actions in draft or approval states when risk is non-trivial.
- Operations briefing (optional) — Summarize sources into concise, owner-aware, non-invented updates.

Self-pace this loop. After each iteration, run the check command, read the output, and only continue if the exit condition is not met. Stop when the exit condition passes or max iterations is reached. Give a short status update each pass.

Paste the kickoff prompt into Cursor, Claude Code, or Codex. Deeplinks do not install hook files.

Steps

1. Intake documents

Collect uploads from portal, email, or integrations; track cases.

2. Extract structured data

AI reads documents with per-field confidence scores.

3. Build master record

Merge fields; resolve conflicts across source documents.

4. Human review

Route low-confidence or judgment calls for sign-off.

5. Deliver outputs

Fill templates, export packages, or push to downstream systems.

Flow diagram

Guardrails

Rules the agent must follow so it cannot cheat the exit condition.

  • Require human approval before customer-facing sends, payments, or legal submissions unless pre-approved templates apply
  • Preserve full audit trail linking source data to every automated action
  • Escalate compliance, safety, or regulatory-sensitive items immediately

More Operations loops