watchers

Run watch_rss.py with a --name and --url argument pointing at the RSS or Atom feed. The first run records a baseline silently; subsequent runs print only new entries as Markdown, deduplicated by guid or id.

Use watch_github.py with --repo owner/name and --scope set to issues, pulls, releases, or commits. Set GITHUB_TOKEN in the environment to avoid the 60 requests per hour anonymous rate limit.

Yes, watch_http_json.py polls any JSON endpoint returning a list of objects. Configure the dedup field with --id-field, locate nested lists with --items-path like data.events, and pass custom headers with repeatable --header flags.

The first run records a baseline of all current item IDs without emitting anything, preventing replay of the entire existing feed. To force a replay, delete the state file in the watcher-state directory before the next run.

The shared watermark helper caps the stored ID set at 500 entries by default, discarding the oldest IDs. You can adjust the max_seen limit when instantiating the Watermark class in a custom script.

No, all three watcher scripts use only the Python standard library: urllib for HTTP, xml.etree.ElementTree for feed parsing, and json for state. No pip installs are required.