discover-changelog

Discover and Set Up Changelog Sources

Given a company name or domain, find their changelog/release-note sources, add them to Releases, and fetch recent entries. This is the manual, Claude-driven onboarding flow — it uses the releases CLI against the production API and does not dispatch to managed agents. (If you want managed-agent onboarding, use releases admin discovery onboard <company> directly.)

Arguments

• $ARGUMENTS — Company name, domain, or org slug (e.g., "Vercel", "linear.app", "stripe")

Prerequisites

• releases CLI installed and authenticated with an admin API key (releases whoami to verify).
• Invoke the releases:finding-changelogs skill before evaluating URLs — it's the source of truth for priority order (well-known files > link relations > feeds > GitHub > raw markdown > scraping) and content-verification red flags.
• Invoke releases:managing-sources before writing — it documents the tool reference table, naming conventions, and the validation workflow.

Process

Step 1: Identify the Domain

If given a company name rather than a domain, determine the primary domain. Check if an org already exists:

releases admin org list --json | jq '.items[] | select(.name | test("<query>"; "i"))'

If the org exists and has a domain, use it. Otherwise, infer the domain from the company name.

Some companies use multiple TLDs or redirect between domains (e.g., claude.ai vs claude.com, with support on support.claude.com). If the primary domain is sparse (login page, app, or redirect), check alternate TLDs and the parent company domain too.

Step 2: Manual Discovery

With local-only discovery retired, Claude does the discovery work directly — following the priority order in releases:finding-changelogs. Work through these tiers until you have candidate URLs:

1. Well-known files — curl -sI https://<domain>/.well-known/changelog.json, then releases.json, changelog.txt, /AGENTS.md, /changelog.md, /releases.md (and uppercase variants).
2. Link relations on the homepage — fetch the homepage and grep for rel="alternate" feeds or rel="changelog" link tags.
3. Footer and nav scan — curl -sL https://<domain> | grep -i -oE 'href="[^"]*(changelog|release|whats-new|updates)[^"]*"'. Footers often link changelogs that aren't in the main nav.
4. Subdomains — docs.<domain>, support.<domain>, help.<domain>, developers.<domain>, status.<domain>, <product>.docs.<domain>. curl -sI to check each resolves (200, not a redirect to the homepage).
5. Provider-specific hints — if the page is hosted on a known changelog provider (Mintlify, ReadMe, Intercom, Zendesk, Docusaurus, WordPress, Ghost), the finding-changelogs skill lists their feed-URL conventions.
6. GitHub — if the company has a GitHub handle, check repos for active tagged releases:

   curl -sH "Authorization: Bearer $GITHUB_TOKEN" \
     "https://api.github.com/orgs/<handle>/repos?per_page=10&sort=updated" | \
     jq '.[].full_name'