Agent & workflow ready
JSON output, deterministic exit codes, and a raw hscli api hatch let agents and pipelines drive Help Scout directly.
A fast, scriptable CLI for Help Scout — built for terminals, CI pipelines, and AI agents like Claude Code and Codex. Every command speaks JSON and returns clean exit codes.
Why hscli
Four design decisions that make hscli safe to hand to a script — or an agent.
JSON output, deterministic exit codes, and a raw hscli api hatch let agents and pipelines drive Help Scout directly.
Every command supports --output table|json|yaml|csv, plus --jq and --fields for inline filtering and projection.
OAuth tokens live only in your OS keychain — never written to disk in plaintext. hscli api is locked to the Help Scout host.
hscli backup dumps your whole account to JSON with incremental refresh, resume, deletion detection, and attachments.
What you can actually do
hscli composes with the tools you already use — jq, xargs, cron, and your CI runner.
Pipe active conversations into your model and act on its decisions — no integration code.
Incremental by default, resumable, and deletion-aware. Drop it in a cron job or a GitHub Action.
Stream any report as JSON straight into a notebook, your BI tool, or a jq pipeline.
Run hscli as a Model Context Protocol server and Claude — Desktop, Code, or any MCP client — gets one typed tool per command. Reads by default; --allow-writes lets the agent act.
Point a model at last week's inbox, let it spot the most-asked question, and publish the answer straight to your Docs knowledge base — version-controlled and shipped from CI.
Docs knowledge base →Install
npm is the source of truth — Homebrew, Scoop, and Docker track every release.
Then authenticate and run hscli conv list — or see the installation guide.
Install once. Script everything. Hand the rest to an agent.