inbox-triage

Routes ACE inbox emails to opportunity runs and classifies sender tiers for approval-gated replies.

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

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Manually routing inbound ACE email threads to their corresponding opportunity runs, classifying sender permissions, and drafting replies wastes operational time and risks missing urgent messages or sending unapproved communications to external partners and internal teams.

Core Features & Use Cases

  • Per-sender thread isolation and routing: Automatically matches each unread email thread to the correct ACE opportunity run using comms-log thread IDs or subject line opportunity slug conventions, eliminating manual relay of replies.
  • Tiered sender classification: Enforces strict act/correspond/unknown tier rules to determine allowed actions for each sender, preventing unauthorized run mutations, spam-driven actions, or unapproved outbound sends.
  • Intelligent noise filtering: Auto-dismisses system notifications (Google Workspace alerts, ACE-generated opportunity updates) to cut through inbox clutter, while surfacing orphaned or unexpected messages that require human attention.
  • Approval-gated action workflow: Drafts contextually appropriate replies, run advances, or escalations for human review before any outbound send, with full interaction logging to the run's comms-log for stateless memory.
  • Use Case: When an external LLO partner replies to a UAT invitation, the skill routes the thread to the correct opportunity run, classifies the sender as a correspond-tier counterpart, drafts a warm, self-identifying ACE reply for approval, and logs the interaction to the run's records.

Quick Start

Use the inbox-triage skill to process all unread emails in ACE's inbox, route each thread to its corresponding opportunity run, classify sender tiers, propose appropriate actions for human approval, and mark fully handled threads as read.

Frequently Asked Questions about inbox-triage

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

FAQPage Schema
How do I automate email triage for routing inbound Gmail threads to specific workflow runs?

Inbox triage automation processes unread Gmail messages and matches email threads to their corresponding opportunity runs using comms-log thread IDs or subject line opportunity slugs. This eliminates manual relay by automatically isolating and routing each sender's thread to the correct run.

How does sender classification prevent unauthorized replies or run mutations in inbox automation?

Sender classification enforces tiered access rules by categorizing senders as internal act, external correspond, or unknown. This prevents unauthorized run mutations, blocks spam-driven actions, and ensures unapproved outbound sends to external partners are stopped before execution.

How do I filter system notifications and workspace alerts to reduce manual inbox noise filtering?

Noise filtering automatically dismisses system-generated notifications like Google Workspace alerts and opportunity updates. This cuts through inbox clutter while surfacing orphaned or unexpected messages that require human attention, significantly reducing manual triage effort.

Can I draft approval-gated replies for external partners without risking unapproved outbound sends?

Yes, an approval-gated action workflow drafts contextually appropriate replies, run advances, or escalations for human review before any outbound send. All interactions are logged to the run's comms-log, providing stateless memory for future reference.

Does this inbox automation tool support tracking open-thread aging for external counterparts?

Yes, the tool tracks open-thread aging for external counterparts alongside unread message processing. It matches threads to runs, classifies senders, proposes actions for approval, and marks fully handled threads as read to maintain clean inbox operations.