topic-memory

Manages the living content-topics.json — CRUD operations on topics, scoring history, revisit scheduling, performance data integration, and archive management. Use when reading, adding, updating, or archiving topics in the content intelligence system.

Model	Source
sonnet	pack: content-pumper

Full Reference

┏━ 🧠 topic-memory ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓ ┃ Living state layer for the content intelligence ┃ ┃ system — reads, writes, archives, and scores ┃ ┗━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┛

topic-memory

Manages content-topics.json — the single source of truth for the Topic Brain system. All other skills read from or write to this file via the operations below. Idempotent — safe to call repeatedly.

---

JSON Schema

{
  "version": "1.0",
  "lastUpdated": "<ISO timestamp>",
  "config": {
    "autoThreshold": 80,
    "checkIntervalHours": 6,
    "maxQueueSize": 50,
    "scoringWeights": {
      "freshness": 0.20,
      "searchVolume": 0.20,
      "socialBuzz": 0.15,
      "polarizationHeat": 0.10,
      "brandFit": 0.20,
      "competitionGap": 0.15
    }
  },
  "topics": [
    {
      "id": "<uuid>",
      "title": "<string>",
      "slug": "<string>",
      "category": "<string>",
      "verticals": ["<string>"],
      "score": 0,
      "signals": {
        "freshness": 0,
        "searchVolume": 0,
        "socialBuzz": 0,
        "trendVelocity": "rising | stable | falling",
        "polarizationHeat": 0,
        "brandFit": 0,
        "competitionGap": 0
      },
      "sentiment": {
        "polarized": false,
        "sides": [
          {
            "label": "<string>",
            "position": "<string>",
            "estimatedSize": 0,
            "triggers": ["<string>"],
            "engagement": 0
          }
        ]
      },
      "history": {
        "firstSeen": "<ISO timestamp>",
        "lastScored": "<ISO timestamp>",
        "lastWritten": "<ISO timestamp | null>",
        "timesWritten": 0,
        "nextCheck": "<ISO timestamp>",
        "articles": [
          {
            "url": "<string>",
            "publishedAt": "<ISO timestamp>",
            "site": "<string>",
            "wordCount": 0
          }
        ],
        "performance": [
          {
            "recordedAt": "<ISO timestamp>",
            "source": "ga4 | gsc",
            "sessions": 0,
            "clicks": 0,
            "impressions": 0,
            "avgPosition": 0,
            "ctr": 0
          }
        ]
      },
      "status": "discovered | queued | writing | published | monitoring | archived",
      "targetMode": "evergreen | trending | seasonal",
      "targetSites": ["<string>"]
    }
  ],
  "archive": []
}

Archive entries use the same structure as topics.

---

Operations

Operation	Input	Action
add-topic	title, category, verticals, targetMode, targetSites	Append to topics[] with uuid, default signals (all 0), status=discovered
update-score	topicId	Recalculate composite score from signals using config.scoringWeights
update-signals	topicId, signals patch	Merge signal values, then call update-score
update-sentiment	topicId, sentiment object	Replace sentiment block, set lastScored
record-article	topicId, url, site, wordCount	Push to history.articles, increment timesWritten, set lastWritten, call set-next-check
record-performance	topicId, source, metrics	Push to history.performance[]
set-next-check	topicId	Set nextCheck = now + config.checkIntervalHours
archive-topic	topicId	Move from topics[] to archive[], set status=archived
get-leaderboard	—	Return topics[] sorted by score descending
get-due-for-check	—	Return topics[] where nextCheck < now

---

Status Lifecycle

discovered
  └─▸ queued        (score ≥ autoThreshold, within maxQueueSize)
       └─▸ writing  (assigned to content-pumper)
            └─▸ published  (article recorded in history.articles)
                 └─▸ monitoring  (tracking performance post-publish)
                      └─▸ archived  (auto-archive rules triggered)

---

Scoring Formula

Normalize each signal to 0–100 before weighting:

composite = Σ(normalized_signal × config.scoringWeights[signal_name])

Signals mapped to weights:

• freshness × 0.20
• searchVolume × 0.20
• socialBuzz × 0.15
• polarizationHeat × 0.10
• brandFit × 0.20
• competitionGap × 0.15

trendVelocity is categorical (rising | stable | falling) — not included in the weighted sum but used in auto-archive rules and tie-breaking.

---

Auto-Archive Rules

Archive a topic when ALL three conditions are true:

• history.timesWritten > 3
• signals.trendVelocity === "falling"
• signals.socialBuzz < 0.2

Call archive-topic for any topic matching all three. Log the trigger in a note field on the archived entry.

---

File Location

Default path: content-topics.json at project root.

Override via env var CONTENT_TOPICS_PATH. Always read the full file, mutate in memory, write back atomically. Never partial-write.

---

Integration

Skill	Reads	Writes
trend-scanner	config.checkIntervalHours	update-signals, add-topic
topic-scorer	topics[], config.scoringWeights	update-score
sentiment-mapper	topics[]	update-sentiment
content-pumper-pimp	get-leaderboard, get-due-for-check	record-article, status transitions
content-scheduler	get-leaderboard	set-next-check