Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when the user asks to "find influencers", "build an influencer list", or "discover creators in [niche]"; produces a multi-platform candidate pool, per-influencer profiles with audience and engagement metrics, authenticity red-flag screening, and a tiered shortlist with fit scores. Not for scoring or ranking a known shortlist — use fit-scorer.
.claude/skills/aiskillstore-influencer-discovery/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 24% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 26% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 26% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 85% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 74% | 0% |
Find the right influencers for your brand by searching across platforms, screening for audience fit and authenticity, and building a tiered candidate list ready for scoring.
Find 20 influencers in [niche] for [brand/product]Find influencers in [niche] with 50K-200K followers on TikTok and Instagram,
based in [location], engagement above 4%, who have worked with brands like [brand]entity-optimizer brand profile and any audience-mapper output if present in memory; existing roster records under memory/creators/ (dedupe the candidate pool against creators already rostered by creator-registry).memory/influencer/influencer-discovery/YYYY-MM-DD-<topic>.md — search criteria, candidate pool stats, per-influencer profiles, tiered shortlist with fit scores. Roster-worthy shortlisted creators (verified handles, contact path, audience stats) go as one-line updates to memory/events/creators.ndjson via an authorized operation: propose request to registry-events.py — only creator-registry writes canonical records under memory/creators/.memory/hot-cache.md.> Emit the standard shape from skill-contract.md §Handoff Summary Format.
This family has no live integrations required (Tier 1): the skill works with only the inputs the user provides. Ask the user for niche, platforms, follower band, engagement floor, location, and exclusions, then reason over what they supply plus any public handles they share.
Where a tool could sharpen results, use ~~ connector placeholders:
~~influencer database — bulk discovery, follower/engagement metrics, audience demographics.~~social platform analytics — native creator-marketplace data, trending sounds, related accounts.~~CRM — import the shortlist and dedupe against existing partners.~~audience overlap — estimate creator-audience vs. brand-audience match.Keyless candidate-card metadata (oEmbed): YouTube (https://www.youtube.com/oembed?url=<video-url>&format=json), TikTok (https://www.tiktok.com/oembed?url=<post-url>), and X (https://publish.twitter.com/oembed?url=<post-url>) return a post's title, author name/handle, and thumbnail with no key — enough to auto-fill a candidate's profile row from pasted links instead of hand-copying. Metadata only: no follower or engagement metrics, so those stay ~~influencer database or manual export — except YouTube, below.
Measured YouTube metrics (free key): python3 "${CLAUDE_PLUGIN_ROOT}/scripts/connectors/youtube.py" channel @handle returns the real displayed subscriber count, total views, and video count, and youtube.py videos @handle --limit 10 adds per-video views/likes/comments — upgrading a YouTube candidate's profile row from Estimated to Measured. Free YOUTUBE_API_KEY (10,000 units/day; one channel check ≈ 1–3 units). ToS boundary: vet a named shortlist, don't build a bulk creator database — quota extensions are refused for competitive harvesting. See scripts/connectors/README.md.
See CONNECTORS.md for the free/keyless recipe per category and the opt-in MCP layer. None are required — every step degrades to user-supplied inputs.
Each step has a fill-in block in references/templates.md — copy the matching block. This skill does not compute final fit scores; the per-influencer score in step 4 is a triage signal that fit-scorer refines downstream.
Save the report to memory/influencer/influencer-discovery/YYYY-MM-DD-<topic>.md and, with permission, cache the active shortlist in working memory. Submit each roster-worthy creator as an authorized operation: propose request through registry-events.py to memory/events/creators.ndjson, then recommend creator-registry for resolution; proposal count never changes authority.
User: "Find 15 micro-influencers (10K-100K followers) in sustainable fashion for a new eco clothing brand."
Output: 43 candidates surfaced, 15 pass all filters with fit scores above 18/25. Top pick @sustainablestyle_sarah (47K IG + 23K TikTok, 5.2% ER, prior eco-brand partners) scores 24/25; shortlist tiered into 5 high-engagement leads, 7 mid-tier, 3 rising stars; report saved and top handles promoted. Full walkthrough in references/templates.md.
Primary: fit-scorer — score and rank the discovered candidates with weighted criteria before outreach.
Alternates (same influencer family):
Termination: Maintain a visited-set. If a skill has already been invoked this session, stop and report chain-complete rather than re-invoking it. Max chain depth is 3 hops from the originating request; stop and summarize when reached.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 32,091 | 38,007 | +18% | 1 | 1 | 0% | 6,230 | 8,451 | +36% | 0 | 0 | — |
case-02 | fail→pass | 35,711 | 31,089 | -13% | 1 | 1 | 0% | 6,226 | 7,740 | +24% | 0 | 0 | — |
case-03 | fail→fail | 36,371 | 35,077 | -4% | 1 | 1 | 0% | 6,226 | 8,447 | +36% | 0 | 0 | — |
case-04 | fail→pass | 15,802 | 9,073 | -43% | 1 | 1 | 0% | 2,939 | 3,706 | +26% | 0 | 0 | — |
case-05 | fail→fail | 30,016 | 15,621 | -48% | 1 | 1 | 0% | 5,006 | 4,674 | -7% | 0 | 0 | — |
case-06 | fail→fail | 26,374 | 17,135 | -35% | 1 | 1 | 0% | 3,966 | 4,951 | +25% | 0 | 0 | — |
case-07 | fail→fail | 18,022 | 7,374 | -59% | 1 | 1 | 0% | 3,301 | 3,569 | +8% | 0 | 0 | — |
case-08 | fail→pass | 14,446 | 6,160 | -57% | 1 | 1 | 0% | 2,572 | 3,229 | +26% | 0 | 0 | — |
case-09 | pass→pass | 13,051 | 6,509 | -50% | 1 | 1 | 0% | 1,978 | 3,299 | +67% | 0 | 0 | — |
case-10 | fail→pass | 12,209 | 10,296 | -16% | 1 | 1 | 0% | 2,252 | 4,177 | +85% | 0 | 0 | — |
case-11 | fail→pass | 14,278 | 9,123 | -36% | 1 | 1 | 0% | 2,144 | 3,723 | +74% | 0 | 0 | — |
case-12 | fail→pass | 13,311 | 7,000 | -47% | 1 | 1 | 0% | 1,885 | 3,267 | +73% | 0 | 0 | — |
case-13 | fail→pass | 10,371 | 4,858 | -53% | 1 | 1 | 0% | 1,531 | 2,904 | +90% | 0 | 0 | — |
case-14 | fail→pass | 16,004 | 5,662 | -65% | 1 | 1 | 0% | 2,386 | 3,080 | +29% | 0 | 0 | — |
case-15 | fail→pass | 9,351 | 4,910 | -47% | 1 | 1 | 0% | 1,279 | 3,006 | +135% | 0 | 0 | — |
case-16 | fail→pass | 22,855 | 3,586 | -84% | 1 | 1 | 0% | 3,165 | 2,763 | -13% | 0 | 0 | — |
case-17 | pass→pass | 13,288 | 8,483 | -36% | 1 | 1 | 0% | 2,104 | 3,594 | +71% | 0 | 0 | — |
case-18 | fail→pass | 17,138 | 11,658 | -32% | 1 | 1 | 0% | 2,429 | 3,908 | +61% | 0 | 0 | — |
case-19 | fail→pass | 12,645 | 3,771 | -70% | 1 | 1 | 0% | 1,880 | 2,821 | +50% | 0 | 0 | — |
case-20 | pass→pass | 8,889 | 2,230 | -75% | 1 | 1 | 0% | 1,259 | 2,541 | +102% | 0 | 0 | — |
case-21 | fail→pass | 7,332 | 3,817 | -48% | 1 | 1 | 0% | 1,133 | 2,859 | +152% | 0 | 0 | — |
case-22 | fail→pass | 14,237 | 9,346 | -34% | 1 | 1 | 0% | 2,008 | 3,496 | +74% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +64 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.