From syndicate-pipeline
Run the T1–T4 dedup pipeline to cluster near-duplicate items across the rolling window.
How this skill is triggered — by the user, by Claude, or both
Slash command
/syndicate-pipeline:syndicate-dedup [--window-days 10] [--tiers 1,2,3,4]When to use
When the user asks to "run dedup", "cluster items", "merge duplicates", or after ingesting new content that should join existing clusters.
[--window-days 10] [--tiers 1,2,3,4]This skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
You are running dedup. **Writes to the DB** — sets `cluster_id`,
You are running dedup. Writes to the DB — sets cluster_id,
is_primary, and cluster_method. Idempotent.
Defaults: window=10 days, tiers=1,2,3,4 (T1 exact URL, T2 fuzzy title,
T3 simhash, T4 semantic embedding). Parse $ARGUMENTS for overrides:
cd "${SYNDICATE_REPO:-$(pwd)}" && uv run python -m pipeline.cli dedup --window ${window_days:-10} --tiers ${tiers:-1,2,3,4}
Output envelope:
{"ok": true, "result": {"examined": N, "new_clusters": N, "items_demoted": N, "matched_phase1": N, "matched_phase2": N, "method_counts": {"singleton": N, "t1_exact": N, "t2_fuzzy": N, "t3_simhash": N, "t4_semantic": N}, "errors": []}, "log_path": "..."}
✓ dedup · examined=X clusters=Y demoted=Z [singleton=A t4_semantic=B ...]
items_demoted is the count of rows that lost is_primary=1 because a
better representative was chosen for their cluster. Demoted items are
excluded from the FE feed but kept in the DB for reaction-graph integrity.
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.
npx claudepluginhub aadarshvelu/syndicate --plugin syndicate-pipeline