Content Curate

Content Curation

Overview

Present stage:mined content queue issues as a batch for quick human decisions: approve, skip, merge, or edit.

Process

1. Fetch mined items:

   gh issue list --repo jonesrussell/jonesrussell --label "stage:mined" --json number,title,body,createdAt --limit 20
   

2. Sort mined items by confidence score (highest first). Extract the confidence value from the **Confidence:** line in each issue body. Items without a confidence value sort last (legacy items from before grouping was implemented).

3. Validate each mined item. For each fetched issue, extract the structured seed data from the issue body and validate against the mined-seed schema:

   node ~/dev/blog/schemas/validate.js mined-seed /tmp/mined-seed-<issue-number>.json
   

   • If validation passes, include in the curation batch.
   • If validation fails, flag it to the user: "Issue #N has malformed seed data: . Skipping — fix the issue manually or re-mine."
   • Do not silently process malformed items.

4. Identify the item source and present accordingly. Two distinct item types exist:

   Git-activity items (from waaseyaa/framework, waaseyaa/giiken, etc.):

   • Show: issue number, title, confidence score, source repo
   • Show: commit count (count comma-separated entries in the Commits: line). For items with multiple commits (grouped seeds), show the commit count rather than individual SHAs. The user does not need to see every SHA during curation.
   • Show: content seed (what was built/changed), suggested type and channels
   • Recommendation: why this makes a good post or why not
   • Path: could become blog post OR standalone text-post

   North Cloud items (label: source:north-cloud):

   • Show with a [NC] prefix so they're visually distinct
   • Show: article title as a link, source domain, quality score, topics
   • Recommendation: whether the article is worth amplifying to your audience
   • Path: always text-post (sharing, not writing) — no blog post
   • Frame the decision as: "Is this worth your audience's attention?"

5. Ask for a decision on each item:

   • Approve — confirm or adjust channels, then build a curated-item JSON and validate:

     cat > /tmp/curated-item-<N>.json << 'CURATED'
     {
       "source": "<from mined seed>",
       "source_ref": "<from mined seed>",
       "content_seed": "<from mined seed, possibly edited>",
       "suggested_type": "<from mined seed>",
       "suggested_channels": <from mined seed>,
       "mined_at": "<from mined seed>",
       "curated_type": "<confirmed type>",
       "curated_channels": ["<confirmed channels>"],
       "curation_action": "approved",
       "curated_at": "<current ISO 8601 timestamp>",
       "editorial_notes": "<optional: human notes on angle>"
     }
     CURATED
     node ~/dev/blog/schemas/validate.js curated-item /tmp/curated-item-<N>.json
     

     If validation fails, show the error and fix before proceeding. If validation passes, update the labels:

     gh issue edit <N> --repo jonesrussell/jonesrussell --remove-label "stage:mined" --add-label "stage:curated"
     

     After labeling, fetch the source issue for full context, then invoke social-media-posts before writing a single word of copy. This is mandatory — not optional, not "check after." The skill defines platform rules (X char limits, hashtag policy, GitHub link requirements) that cannot be reliably recalled from memory.
   • Skip — close the issue immediately to remove it from the queue. This distinguishes intentional skips from auto-expiry (which uses the auto-expired label).

     gh issue close <N> --repo jonesrussell/jonesrussell --comment "Skipped during curation."
     gh issue edit <N> --repo jonesrussell/jonesrussell --add-label "curate:skipped"
     

   • Merge — combine with another item (ask which one), update the target issue's seed, skip the source
   • Edit — update the seed material, type, or channels before approving

6. Batch summary. After processing all items, report:

   • N approved (ready for production)
   • N skipped
   • N merged
   • Suggest running /content-produce next for the approved items

Presentation Style

Keep it scannable. One item at a time.

Git-activity item:

**#15: [content] SovereigntyProfile shipped in Waaseyaa** (confidence: 0.82)
Source: waaseyaa/framework · 3 commits (2026-04-04)
Seed: Added SovereigntyProfile to Layer 0. Communities declare local/hybrid/cloud sovereignty mode.
Suggested: text-post → x, linkedin, facebook

Good candidate. Concrete feature, ties to data sovereignty narrative.

→ Approve / Skip / Merge / Edit?

North Cloud item:

[NC] **#47: [content] Why your PHP app is slower than it needs to be**
Source: North Cloud · coforge:core · dev.to (quality: 78)
Article: https://dev.to/example/why-your-php-app-is-slower
Topics: php, performance, optimization

Worth sharing. Practical, audience-relevant, no paywall. Good for a quick reshare with a take.

→ Share / Skip / Edit?

When Queue Is Empty

If no stage:mined items exist, say so and suggest:

• Running the mining skill (/content-mine) to scan recent activity
• Creating a content idea manually via the issue template