Pull Analytics X (Analytics Agent — X pipeline)

X's sibling to /pull-analytics. Same process, scoped to platform: x notes and BUFFER_CHANNEL_ID_X.

Important constraints

Same as /pull-analytics: post-metrics is experimental on Buffer's side, refreshed once daily, can lag ~24h. A null/missing metric is never a zero — report it as not-yet-available.

Required environment

BUFFER_ACCESS_TOKEN (shared with the LinkedIn/X scheduler). Stop and say so plainly if missing.

Process

1. Find posts to check

Every platform: x note in Scheduled/ or Published-Posts/ with a real buffer_post_id and a scheduled_date at least 1 day in the past.

2. Query Buffer

Same query shape as /pull-analytics:

query {
  post(input: { id: "<buffer_post_id>" }) {
    id
    metrics { type name value unit }
  }
}

Map to the Analytics table columns: impressions, reach, reactions (X's likes), comments (X's replies), shares (X's reposts), clicks, engagementRate. If a metric Buffer reports for X doesn't map cleanly to these LinkedIn-shaped columns, report the raw metric name/value in ## Publishing Notes rather than forcing it into the wrong column.

3. Update publish status

Same move-to-Published-Posts/ logic as /pull-analytics, keyed off platform: x notes.

4. Append the snapshot

Create/update Analytics/<id>.md from _Templates/Analytics-Record.md with platform: x, appending — never overwriting — a Snapshots row.

5. Flag standout performance

Same rule as /pull-analytics: compare against this post's own prior snapshots or other X posts specifically (not LinkedIn's baseline — the platforms don't share a baseline). Say plainly if there isn't yet enough X data for a meaningful comparison.

Notifications

Same triggers as /pull-analytics: a post failed to go out, or genuine outperformance against a real X-specific baseline.

Hard rules

Identical to /pull-analytics's hard rules, platform-scoped to X.