decisions-sync

Sync human-edited decision values from Google Docs into YAML configuration files.

1|2|Updated Apr 1, 2026
One-click install
npx skills add https://github.com/dimagi-internal/ace --skill decisions-sync
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: decisions-sync
Source: https://github.com/dimagi-internal/ace/tree/main/skills/decisions-sync
Command: npx skills add https://github.com/dimagi-internal/ace --skill decisions-sync

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

When operators edit the human-readable decisions Google Doc for an ACE Connect opportunity run, those changes are not automatically reflected in the structured decisions.yaml file that drives subsequent lifecycle phases. This skill bridges that gap to ensure human overrides are honored in future runs.

Core Features & Use Cases

  • Human Edit Sync: Diffs the per-run decisions.gdoc against the existing decisions.yaml to identify all human-made changes to decision values.
  • Preserved Audit Trail: Keeps original AI-generated default values intact, marks overridden entries with a status and override value, and appends new considered options from the Google Doc.
  • Clear Change Reporting: Generates a concise summary of all overrides, added options, and unmatched rows so operators can verify changes at a glance.
  • Use Case: After adjusting the target beneficiary count or AI confidence threshold in the run's decisions Google Doc, use this skill to sync the changes so the next ACE run uses your updated parameters instead of the original AI defaults.

Quick Start

Use the decisions-sync skill to sync all human-edited decisions from the per-run Google Doc into the corresponding decisions.yaml file for your current ACE opportunity run.

Frequently Asked Questions about decisions-sync

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I sync human-edited decisions from a Google Doc to a YAML file?

To sync human-edited decisions from a Google Doc to a YAML file, this skill diffs the per-run decisions document against the existing decisions.yaml to identify human-made changes. It then updates the structured YAML configuration while preserving original AI defaults and appending new considered options.

How does ACE Connect preserve AI default values when operators override decisions?

ACE Connect preserves AI default values by keeping original entries intact when human overrides are applied. The sync process marks overridden entries with a status and override value, and appends new considered options from the Google Doc to maintain full auditability of all decision modifications.

Can I update run management parameters like beneficiary count in decisions.yaml after editing the Google Doc?

Yes, you can update run management parameters like beneficiary count or AI confidence thresholds in decisions.yaml after editing the Google Doc. The skill synchronizes these human-edited values so subsequent ACE opportunity runs use the updated parameters instead of original AI defaults.

What is the best way to track human overrides and added options in structured run configuration?

The best way to track human overrides in structured run configuration is to diff the human-readable decisions document against the existing YAML. This generates a concise summary of all overrides, added options, and unmatched rows so operators can verify changes at a glance.

Why are my Google Doc decision edits not reflected in the structured run state for subsequent lifecycle phases?

Google Doc decision edits are not automatically reflected in the structured run state because the human-readable document and the decisions.yaml file driving lifecycle phases are separate. This skill bridges that gap to ensure human overrides are honored in future runs.

Does decisions-sync work with per-run ACE Connect opportunity runs without additional dependencies?

Yes, decisions-sync works with per-run ACE Connect opportunity runs without additional dependencies. It requires only the per-run decisions Google Doc and the existing decisions.yaml file to synchronize human-edited decision values into the structured run state.