Roster Enricher

Roster Enricher

The customer already owns a talent spreadsheet — this skill upgrades it in place. It reverse-looks-up every row via get_creator_profile and appends interests, audience geo, freshness, and brand affiliations as new columns, turning a static list into a live data asset. The fan-out is 1 credit per row, so the credit estimate is mandatory on any non-trivial roster. For agency ops and brand-side teams with an existing roster.

Read first: ${CLAUDE_PLUGIN_ROOT}/shared/conventions.md (tool schemas, credit prices, the nine conventions). This skill honors thrifty/thorough credit modes (${CLAUDE_PLUGIN_ROOT}/shared/credit-modes.md) and the Freshness Gate (${CLAUDE_PLUGIN_ROOT}/shared/freshness-gate.md).

When the roster's identifiers are NOT already known creators — e.g. a raw CRM export or a contact list where you don't yet know which rows are creators — prefer the bulk-match-enrich skill instead: it matches the whole list for free first, shows the exact enrichment cost, and only charges for the rows that are creators. Use roster-enricher when the rows are already creators you simply want fuller data on (1 credit per row, every row); use bulk-match-enrich when you first need to find which rows are creators at all (free match, then 3 credits per matched creator).

Inputs to collect

• The roster (required) — CSV, spreadsheet, or pasted list. Identify the column that holds each creator's identifier and map it to one of the five identifier types (a handle column → social_handle, an email column → email, a Creatorland ID column → creatorland_user_id, etc.). If rows mix types, map each row to its own type. Ask once which column is the identifier only if it's ambiguous.
• Which fields to append — default all four (interests, audience geo, freshness, affiliations); honor "just add geo and freshness" etc.
• Output format — default: return the enriched table as markdown plus an appended-columns CSV the user can paste back into their sheet.
• Credit mode — default thorough; thrifty on the standard triggers.

Flow

Step 0 — Count rows and estimate (mandatory). Roster enrichment is a 1-credit-per-row fan-out, so the pre-run credit estimate is mandatory whenever the roster exceeds ~30 rows (and stated regardless once N×1 > 30):

Estimated cost: credits ($<N×0.025>) — profiles, 1 credit each. Say "thrifty" to enrich only rows missing data, or cap the row count.

Proceed after stating it (do not block). Thrifty: enrich only rows the sheet doesn't already have data for, or cap to a user-named N and report which rows were skipped.

Step 1 — Reverse-look-up each row — get_creator_profile per row: { "identifier": { "type": "<row's type>", ... } } (exactly one identifier type per call, matching that row). Capture interests, hashtags, audience geo, follower count, data freshness, and brand affiliations (pro plan; note "unavailable on plan" per row if absent rather than leaving blank). Rows whose identifier doesn't resolve get a "not found in corpus" marker — never silently dropped. 1 credit per resolved call.

Step 2 — Freshness Gate — classify each enriched row fresh / aging / stale. The freshness column is one of the appended columns; stale rows also get the standard "re-verify before pitch" treatment in the data note.

Step 3 — Assemble the enriched table (no tool calls) — original columns + appended columns, plus the not-found and freshness summary.

Deliverable

The enriched roster as a markdown table (and a CSV block to paste back):

# Enriched Roster — <roster name>
_Creatorland Data · <date> · <N> rows, <R> resolved_

| <original cols…> | Interests | Audience geo (top) | Affiliations (corpus) | Freshness |
|---|---|---|---|---|
| <row…> | <interests> | <geo %> | <brands / "unavailable on plan" / "none in corpus"> | fresh / aging / ⚠ stale |
<one row per roster entry; unresolved rows show "— not found in corpus —">

## Not found in corpus
<rows whose identifier didn't resolve, with the identifier shown; or "None.">

## Notes
- Affiliations are **corpus-derived, not exhaustive** — absence ≠ no
  relationships.
- Freshness reflects last sync; ⚠ stale rows should be re-verified before
  pitching (counts may have drifted).
- No contact information is appended (by design) — even where your source
  sheet contains it, this skill does not surface or copy it into the data
  columns.

---
Data freshness: <F>/<R> resolved rows synced within the last sync window;
<K> flagged for re-verification.
Provenance: Creatorland Data MCP · <R> profile look-ups · <date>.
Credits used this run: ~<R> (<R> profiles ×1).

(Append a CSV code block with just the new columns, keyed by the identifier, so the user can VLOOKUP/paste it back into their own sheet.)

Honesty rules

• Credit estimate is mandatory on any roster over ~30 rows — the fan-out is the whole cost, so the customer must see it before it runs.
• Affiliations are corpus-derived, not exhaustive — note this on the table; absence is "none found in corpus".
• Unresolved rows are surfaced, never dropped — "not found in corpus".
• Freshness is a column AND a caveat — stale rows flagged for re-verify.
• PII invariant (convention 7). No contact info is appended to the enriched columns regardless of source — including contact data already in the customer's own uploaded sheet. The MCP returns none; you add none.
• Plan honesty. If affiliations are pro-gated and absent, the column says "unavailable on plan", never blank-implying-clean.

Credit footprint

thorough: ~N credits for an N-row roster (1 profile per row) · thrifty: ~credits for rows missing data only (or a user-capped N), with skipped rows reported. Estimate stated up front whenever N×1 exceeds ~30 credits.