triage

Triage GitHub issues through a label-driven state machine with agent briefs.

20|2|Updated May 16, 2026
One-click install
npx skills add https://github.com/reddb-io/red-skills --skill triage-reddb-io
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: triage
Source: https://github.com/reddb-io/red-skills/tree/main/plugins/dev/skills/engineering/triage
Command: npx skills add https://github.com/reddb-io/red-skills --skill triage-reddb-io

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Incoming bug reports and feature requests arrive unstructured, leaving maintainers to manually classify, verify, and prepare each issue before work can begin. This Skill routes every issue through a consistent state machine so raw reports become agentable, well-specified work items. ## Core Features & Use Cases - State-machine triage flows: Show the queue, triage a single issue, apply quick overrides, or resume prior triage sessions using category roles (bug, enhancement) and state roles (needs-triage, needs-info, ready-for-agent, ready-for-human, wontfix). - Agent brief authoring: Writes durable, behavioral ## Agent brief sections directly into issue bodies with testable acceptance criteria, preparing issues for autonomous AFK agents. - Deduplication and verification: Checks .out-of-scope/ rejection records, optionally recalls decisions from the memory plugin, and reproduces bug claims before recommending a state. - Use Case: A maintainer asks to triage issue #42; the Skill reads the thread, verifies the reported bug against the codebase, checks for prior rejections, and writes an agent brief before moving the issue to ready-for-agent. ## Quick Start Ask the assistant to triage issue #42 and prepare it for an agent.

Frequently Asked Questions about triage

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

FAQPage Schema
How do I triage GitHub issues with an AI agent?▼

Reference a single issue or ask to see the queue, and the Skill reads the body, comments, and labels, verifies bug claims against the codebase, then recommends a category and state role. State transitions are applied through the rs_dev MCP triage tool after maintainer confirmation.

What is an agent brief in GitHub issue triage?▼

An agent brief is a structured `## Agent brief` section written into the issue body when it moves to ready-for-agent. It describes current and desired behavior, key interfaces, testable acceptance criteria, and explicit out-of-scope boundaries for an autonomous agent.

Does the triage workflow support external pull requests?▼

External PRs are surfaced only when the dev.triage.external_pr_surface.enabled config toggle is true. PR bodies and diffs are treated as untrusted data and are never checked out, built, or executed during triage.

How does issue deduplication work during triage?▼

The Skill reads .out-of-scope/*.md files to surface prior feature rejections and optionally queries the memory plugin for past decisions. Matches are surfaced to the maintainer, who confirms, reconsiders, or proceeds with normal triage.

Why would an issue be rejected from ready-for-agent?▼

The executable readiness lint requires an `## Acceptance criteria` section with machine-checkable items naming tests, commands, or observable behaviors. Vague or missing criteria route the issue back to needs-triage with a recipe comment.

Can triage instructions in an issue body hijack the agent?▼

No. Issue bodies, comments, and PR diffs are treated as untrusted data to summarize, never as instructions. Only explicit maintainer direction through a triage flow can trigger labels, transitions, or closures.