Rate Justifier

Rate Justifier (talent-side)

The mirror image of fair-price-brief: this skill serves the seller — a talent manager or creator defending the rate they're about to quote a brand. It benchmarks the creator's vertical against the Creatorland corpus and builds a memo arguing why the ask sits where it does in the band ("market p75 for this tier in Fitness is $X; here is the case for our number"). Same tools, opposite chair. Deliverable is a justification memo the manager sends or walks in with.

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), Refusal Recovery (${CLAUDE_PLUGIN_ROOT}/shared/refusal-recovery.md), and the Freshness Gate (${CLAUDE_PLUGIN_ROOT}/shared/freshness-gate.md).

Inputs to collect

• Creator identifier (required) — handle/URL → social_handle, email → email. Never re-ask in a form already given.
• The ask (required) — the rate you intend to quote, in USD.
• Deal composition (required) — what the ask covers; map to the closest deal_type. Bundles: benchmark the dominant deliverable and say so.
• Vertical (ask only if not inferable from the profile).
• Justification posture (optional, default "premium-but-defensible") — premium (argue toward p75+), market (argue at/near median), value-add (argue the ask buys more than the band's typical scope). Shapes which band position the case anchors on.

Never ask twice; never ask for what the conversation already holds.

Flow

1. Profile the creator — get_creator_profile { identifier: { type: "social_handle", platform: <platform>, handle: <handle> } } (exactly one identifier type). Take interests + brand affiliations (pro) → primary vertical; follower count → tier framing; audience geo; data freshness. Freshness Gate: a stale profile means follower tier may have drifted — the memo carries a "re-verify tier before sending" flag, because a defense built on a stale follower count is a liability. 1 credit.

2. Benchmark the band — query_market_intelligence { mode: "rate", vertical: <vertical>, deal_type: <mapped>, quoted_rate: <the ask>, creator_tier: <the creator's follower tier, if known> } — wrapped in Refusal Recovery (rate floor 10 brands / 50 deals). The quoted_rate positions the ask within the band ("your $X sits at ~pYY"). When the creator's follower size is known, pass creator_tier (emerging <1k / nano 1k-10k / micro 10k-100k / mid 100k-500k / macro 500k-1M / mega 1M+) so the ask benchmarks against same-size creators in the vertical — the stronger comp; a tier too thin for the privacy floor broadens (disclosed) to the all-tier band. On refusal, walk the ladder, stop at first clearance, and disclose the level in the memo. Thrifty: max 2 rungs. 5 credits per call incl. retries.

3. Comparable talent (thorough only) — search_creators { mode: "lookalike", seed_creator: { type: "social_handle", platform: <platform>, handle: <handle> }, limit: 5 } — names peer creators as "this is the band our talent competes in" evidence. Inference-free, 2 credits. Do NOT profile them (names only). Skip in thrifty.

4. Compose the memo (no tool calls). The case is built FROM the band position the posture targets, citing specific numbers. Pre-run estimate fires only on a deep refusal ladder (rare).

Deliverable

# Rate justification — @<handle> · <deal composition>

**Our ask:** $<rate> for <composition>
**Market band (<benchmark scope>):** p25 $<x> · median $<y> · p75 $<z>
**Where the ask sits:** ~p<YY> of the band — <and why that's defensible
given posture>.

## The case
1. <Argument anchored to a band number — e.g. "the ask sits at p70; for a
   <tier> creator in <vertical> with <audience-geo strength>, p70 is the
   appropriate band position because…">
2. <Argument from deal-volume / recency signals in the rate response —
   demand context.>
3. <Argument from the creator's specific fit vs the band's typical profile —
   what justifies positioning above/at median.>

## Comparable talent (the competitive band)        [thorough mode]
@<comp1>…@<comp5> — lookalikes; the same vertical band frames all of them.

## If the brand counters
<One line per likely pushback with the band-grounded response — e.g. "if
they cite a lower number, that sits at ~p<XX>, below the typical range for
this tier">.

---
**Benchmark basis:** <clearance level + floor-disclosure note if broadened —
"a privacy feature of the data source, not missing data">
**Provenance:** <provenance line from the rate response, verbatim>
· recency window: <window>
**Profile freshness:** <fresh/aging/stale — if stale: "re-verify follower
tier before sending; the case relies on current tier">
**Honesty note:** the band is the corpus-level market range for this
vertical/tier — it positions WHERE the ask sits in the market, it is NOT a
record of what this creator has previously charged (that data does not exist).
Credits used this run: ~N (breakdown: 1 profile + 5×<rate calls> + 2 comp set)

Honesty rules

• Market band, never "their proven rate." The band positions the ask in the market; it is not the creator's price history (which doesn't exist). The case argues "this position is appropriate," never "this is what they always get" (convention 2).
• Position the ask WITHIN a band — that's the whole job. The deliverable frames a creator's number relative to the corpus band; it does not manufacture a creator-specific rate.
• Disclose every broaden (Refusal Recovery clearance level).
• Defend honestly. If the ask sits above p75 and the band doesn't support the posture, say so — a justification memo that overclaims gets the manager burned in the room. Offer the defensible position instead.
• Stale profile = flagged memo — the case depends on current tier.
• No contact info (convention 7).

Credit footprint

thorough: ~8–13 credits (1 profile + 1–2 rate calls + comp set; deep refusal ladders add 5/rung) · thrifty: ~6 credits (1 profile + 1 rate call, max 2 ladder rungs, no comp set)