ArchAgents

ArchAgents accelerates post-sales work — discovery, integration, migration, and the long tail of customer-specific code — with agents that run on webhooks, schedules, and API triggers. Import a production-ready sample Solution from this repo (PR review, compliance, security triage, changelogs, and more), or build your own: define behavior in YAML, plug in any model, add custom tools and routines.

Try it now

Paste this into Claude Code, Codex, or any AI coding assistant:

Set up ArchAgents in this repo and deploy a working agent I can test.

1) Read https://docs.archagents.com/llms-full.txt
2) Ask me for my email and any missing credentials
3) Install the CLI: brew install ArchAstro/tools/archagent
4) Run: archagent auth login <my-email> && archagent init
5) Pick the best sample Solution from solutions/ for my use case and import it
6) Show me which AgentTemplate, tool templates, routine templates, and setup actions were imported
7) Summarize how to install the imported template into my app

Or run the quickstart directly:

git clone https://github.com/ArchAstro/archagents.git
cd archagents
./quickstart.sh you@company.com

This imports the Onboarding Q&A Solution so you can install the bundled AgentTemplate from the catalog.

Sample Solutions

Solution	What it does	Trigger
Code Review	Reviews every PR with verified inline comments	PR webhook
Compliance Reviewer	Checks PRs against SOC2 / GDPR / your custom rules	PR webhook
Cross-Org Collab	Privacy-by-construction — field guards block code leaks in shared threads	Thread join
Onboarding Q&A	Answers new-hire questions from your knowledge base	Thread join
Platform Health	Daily repo health, GitHub event alerts, and 5xx digests	Mixed
Release Notes	Drafts weekly changelog from merged PRs	Weekly cron
Security Triage	Scans dependencies for CVEs, auto-fixes simple ones, escalates the rest	Daily cron
Threat Intel	Daily security brief from HN + GitHub Advisories cross-referenced against your stack	Daily cron

Each Solution bundle has its own README and ships a deployable AgentTemplate plus any standalone custom tool and routine templates.

Recipes: agents working together

These templates are designed to compose. Install multiples into the same repo for layered coverage:

PR Review Pipeline — Code Review + Compliance Reviewer on the same repo. One reviews architecture and correctness, the other reviews against your compliance rules. Same webhook, two perspectives.

Security Operations — Threat Intel runs the morning brief, Security Triage scans and triages throughout the day. The threat brief identifies what to watch; the triage agent acts on what's exploitable.

Dev Lifecycle — Code Review catches issues on every PR, Release Notes Bot drafts the weekly changelog from the same PRs once they merge. The review history informs what's worth highlighting in the notes.

Install the CLI

macOS

brew install ArchAstro/tools/archagent

Linux

curl -fsSL https://raw.githubusercontent.com/ArchAstro/archagents/main/install.sh | bash

Windows

irm https://raw.githubusercontent.com/ArchAstro/archagents/main/install.ps1 | iex

Plugins for AI coding assistants

Install the local Claude Code and Codex plugins with the setup command:

archagent setup

To install into the current project instead of your user-level config:

archagent setup --scope project

The archagents plugin bundles CLI auth, agent authoring, script and workflow builders, deployment, chat, config management, and embed.

CI

Sample validation and release workflows use ArchAstro/archastro-github-actions:

• validate-sample-scripts.yml schema-validates agents/ and solutions/ bundles.
• check-sample-version-bumps.yml requires changed samples to bump sample.yaml version:.
• release-samples.yml creates missing <slug>-<version>.tar.gz GitHub Releases and deploys solution samples.

The release-and-deploy job passes ARCHASTRO_CI_ACCESS_TOKEN to the reusable action as the ArchAstro system-user token and sets owner: system so solution samples deploy through the system-owned archastro path.

Docs

• Getting Started — zero to one working agent
• For Coding Agents — machine-friendly setup
• CLI Reference — terminal workflow
• Agents — the full agent model