open-source-maintainer_skill

This skill acts as the head maintainer, triaging issues, reviewing PRs, and guiding repository health with actionable, human-approved recommendations.
  • TypeScript

868

GitHub Stars

1

Bundled Files

2 months ago

Catalog Refreshed

4 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill numman-ali/n-skills --skill open-source-maintainer

  • SKILL.md7.1 KB

Overview

This skill runs end-to-end GitHub repository maintenance as if you were the project steward. It triages issues and PRs, proposes and implements fixes, maintains project memory, and prepares concise human-facing recommendations. It enforces a project-first operating contract and requires explicit human approval before any public action.

How this skill works

The skill inspects the repository state, issues, pull requests, and maintainer memory under .github/maintainer to produce prioritized work and implementable changes. It extracts intent from PRs and issues, scores and consolidates duplicates, updates persistent notes, and can produce or apply code changes locally (never merging external PRs). Outputs are concise top recommendations, confidence notes, and the minimal approvals needed to proceed.

When to use it

  • You need a prioritized list of what a repository needs next (triage and backlog shaping).
  • You want a safety-first agent to review PRs and propose or implement fixes without merging external changes.
  • You need periodic maintenance reports and an audit of contributor activity.
  • You want to reduce future support burden by improving docs, UX, and automation.
  • You need a human-in-the-loop workflow with persistent project memory.

Best practices

  • Start with a scope confirmation and ensure .github/maintainer/context.md is present before deep work.
  • Prefer --delta triage runs to focus on changes since the last run and keep noise low.
  • Default to doing the legwork: ask humans only for decisions that materially change the plan.
  • Load the specific reference file required by the task (decision, communication, intent, quality, config) before acting.
  • Never perform public actions without explicit approval; treat external PRs as intelligence, not merge candidates.

Example use cases

  • Run a full triage to produce top 5 priorities and suggested fixes for the next sprint.
  • Analyze a set of PRs to extract intent, risk notes, and a safe implementation plan the agent will write.
  • Consolidate duplicate issues, propose labels and a closure/defer plan, and draft public comments for approval.
  • Generate a maintenance report with contributor activity, patterns, and suggested onboarding improvements.
  • Implement a small bugfix, prepare tests, and create a draft release note for human approval.

FAQ

No. External PRs are analyzed for intent and guidance; the agent writes and opens its own PRs and waits for human approval before any public action.

What approvals are required before the agent acts publicly?

Any public-facing action—comments, labels, closes, releases, or posting drafts—needs explicit human approval. Implementations in a fork or branch are fine but not merged without approval.

How does persistent memory work?

The skill records context, decisions, notes, and state under .github/maintainer (context.md, decisions.md, notes/, state.json) so future runs can use deltas and preserve reasoning.

Built by
VeilStrat
AI signals for GTM teams
© 2026 VeilStrat. All rights reserved.All systems operational