competitor-changelog-watch
Competitor changelog watch
Catches discussions about a competitor’s releases, changelog entries, and version numbers, classifies each result by intent, and routes the digest to your product team.
What it does
- 1
Step 1
Register one catch per competitor with their product name and release-related terms in the query. Example: octolens AND ("changelog" OR "release" OR "update" OR "v3"). Set classify to true so each result carries an intent label. - 2
Step 2
Set the interval to 6h. Competitor release discussions build over days, not minutes. A slower interval costs fewer units and the digest runs daily anyway. If a competitor ships something major, tighten to 1h for 48 hours. - 3
Step 3
Route to a Doozy list, a webhook, or an email digest. The classification labels each result as announcement, question, complaint, comparison, or recommendation. Your agent sorts the digest by intent: complaints first, then comparisons, then announcements. - 4
Step 4
When a competitor ships a new major version, PATCH /v1/catches/{id} to update the query with the new version string. The catch keeps its history and delivery log. You do not need to delete and recreate it. - 5
Step 5
Combine with the competitor-watch playbook for a complete picture. That playbook catches all mentions; this one narrows to release and changelog discussion. Run both and deduplicate by URL in the digest. - 6
Step 6
This playbook ships as a SKILL.md in the dcouple/discatch-playbooks repo so skills.sh and agent skill directories pick it up.
Sources it watches
Common questions
How is this different from the competitor-watch playbook?
The competitor-watch playbook catches all mentions of a competitor at a broad query. This one narrows to release and changelog discussions by adding terms like "changelog", "release", "update", and specific version strings. It answers a different question: what did they ship and how are people reacting to it.
How many competitors can I watch?
As many as you want. Each competitor is one catch. Catches are not priced per keyword, so adding a competitor costs only the units its firings consume. A catch at 6h that finds no new release discussion costs nothing. Ten competitors at 6h with occasional firings is a small number of units per week.
Why use classify on this playbook?
Without classify, the digest is a list of results with prescores. With classify, each result carries a relevance score, a sentiment, and an intent label. For a changelog digest, the intent label tells you whether people are asking questions about the release, complaining about a regression, or comparing the new features to your product.
What sources should I watch for competitor releases?
Hacker News catches developer discussion and Show HN threads. GitHub catches issues and discussions that reference the release. Discourse catches forum threads. Web and news catch blog posts, press coverage, and the competitor’s own changelog page. Add stackexchange if the competitor has a tag on Stack Overflow.
Can I detect a competitor release automatically?
Not with a catch alone. A catch watches for mentions, not for changes to a specific URL. To detect the release itself, poll the competitor’s changelog page with a separate tool and then create or update the catch query with the new version string. The SKILL.md includes a script that does this.
Does discatch draft a response to the competitor release?
No. discatch catches and hands off. It does not draft, post, or suggest replies. Your product team reads the digest and decides whether and how to respond. If you want your agent to draft talking points, route the results to a Claude Code or Codex command with a drafting prompt.
Last reviewed by Parsa Khazaeepoul, founder, dcouple. Copy from docs/playbooks.md.