sponsor-sleuth
sponsor-sleuth: find the brands paying creators your size, price yourself, and pitch them.
Safety scan
A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.
Badge for your README
Markdown
[](/listings/gabrielebolognese--sponsor-sleuth)HTML
<a href="/listings/gabrielebolognese--sponsor-sleuth"><img src="/api/listings/gabrielebolognese--sponsor-sleuth/badge.svg" alt="findsafeskills safety rating" /></a>Install
Install instructions from the README
As written by the repo's author. findsafeskills has not verified these steps.
Requires [Claude Code](https://claude.com/claude-code) and Python 3.10+. There are no Python packages to install.
**As a plugin:**
/plugin marketplace add gabrielebolognese/sponsor-sleuth /plugin install sponsor-sleuth@sponsor-sleuth
Plugin skills are namespaced, so the command is `/sponsor-sleuth:sponsor-sleuth @yourchannel`. You can also just ask Claude "find sponsors for @yourchannel".
**Or as a plain skill**, which gives you the shorter `/sponsor-sleuth` command:
git clone https://github.com/gabrielebolognese/sponsor-sleuth cp -r sponsor-sleuth/skills/sponsor-sleuth ~/.claude/skills/
Get a YouTube API key (free, about 5 minutes)
1. Open [Google Cloud Console](https://console.cloud.google.com/) and create a project. 2. Go to **APIs & Services > Library** and enable **YouTube Data API v3**. 3. Go to **APIs & Services > Credentials > Create credentials > API key**. Restricting the key to the YouTube Data API v3 is a good idea. 4. Run `/sponsor-sleuth setup` and paste the key when asked, or set the `YOUTUBE_API_KEY` environment variable.
**Optional:** `pip install youtube-transcript-api` turns on ad-read detection in video transcripts (best-effort, see Limits).
What was scanned
The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.
.claude-plugin/marketplace.json · Manifest · 15 lines
- {
- "name": "sponsor-sleuth",
- "description": "sponsor-sleuth: find the brands paying creators your size, price yourself, and pitch them.",
- "owner": {
- "name": "Gabriele Bolognese",
- "url": "https://github.com/gabrielebolognese"
- },
- "plugins": [
- {
- "name": "sponsor-sleuth",
- "source": "./",
- "description": "Find every brand that paid YouTube creators your size this month, with proof links, your rate, and drafted pitches."
- }
- ]
- }
README.md · README · 147 lines
- # sponsor-sleuth
- **Find every brand that paid creators your size this month, and pitch them all in 10 minutes.**
- Brands that sponsor creators leave public evidence everywhere: "this video is sponsored by", promo codes, tracking links, YouTube's paid-promotion flag. sponsor-sleuth is a Claude Code skill that collects that evidence from creators who look like you and turns it into a ranked list of brands that are spending money right now, at your size, in your niche. Every brand comes with a link to the exact video it was found in.
- ```
- /sponsor-sleuth @yourchannel
- ```
- ## What you get
- - **A ranked brand list** with proof links. It's scored on size fit, recency, repeat rate (a brand that comes back to the same creator 3 times is paying for results) and language fit.
- - **The headlines:**
- - "23 brands sponsored creators your size in the last 30 days."
- - "8 brands sponsor creators like you but have never sponsored you."
- - "This brand came back to the same creator 4 times. Pitch them first."
- - **The contact path** for the top brands: creator and affiliate program pages, partnership emails on the brand's own site, the affiliate network they run on, and a LinkedIn search link.
- - **Your rate:** median views of your last 10 long-form videos x the CPM range for your niche. For example, 8,000 views x $15-30 = **$120-240 per integration**. The CPM range starts as an assumption and recalibrates from your own closed deals.
- - **10 drafted pitches**, each with 2 follow-ups, written from the evidence ("you sponsored X and Y last month..."), with a 60-second integration concept and your rate.
- - **A pipeline** (`pipeline.csv`, which opens in Sheets or Excel) that tracks drafted, pitched, replied, negotiating and closed, with follow-up reminders.
- - **A reality check.** If no brand pays creators within 3x of your size, it says so instead of handing you pitches that will be ignored. Instead you get:
- - **A growth target:** the smallest creator in your niche that a brand actually paid, in subscribers and views per video, with the proof video.
- - **The gap** between you and that target.
- - **A 90-day plan**, built from what those creators do differently: upload rate, Shorts mix, video length.
- It switches mode based on your size:
- | Size | Mode | What it hunts |
- |---|---|---|
- | 0-1k | **UGC** | Brands that pay for content rather than reach, plus affiliate programs. The pitch is built on your portfolio, and the rate is per asset. |
- | 1k-50k | **Micro** | Brands sponsoring your lookalikes, plus hybrid deals (a flat fee plus affiliate). |
- | 50k+ | **Conquest** | Brands that sponsor your competitors but never you, renewals, and a rate check against your views. |
- ## Install
- Requires [Claude Code](https://claude.com/claude-code) and Python 3.10+. There are no Python packages to install.
- **As a plugin:**
- ```
- /plugin marketplace add gabrielebolognese/sponsor-sleuth
- /plugin install sponsor-sleuth@sponsor-sleuth
- ```
- Plugin skills are namespaced, so the command is `/sponsor-sleuth:sponsor-sleuth @yourchannel`. You can also just ask Claude "find sponsors for @yourchannel".
- **Or as a plain skill**, which gives you the shorter `/sponsor-sleuth` command:
- ```bash
- git clone https://github.com/gabrielebolognese/sponsor-sleuth
- cp -r sponsor-sleuth/skills/sponsor-sleuth ~/.claude/skills/
- ```
- ### Get a YouTube API key (free, about 5 minutes)
- 1. Open [Google Cloud Console](https://console.cloud.google.com/) and create a project.
- 2. Go to **APIs & Services > Library** and enable **YouTube Data API v3**.
- 3. Go to **APIs & Services > Credentials > Create credentials > API key**. Restricting the key to the YouTube Data API v3 is a good idea.
- 4. Run `/sponsor-sleuth setup` and paste the key when asked, or set the `YOUTUBE_API_KEY` environment variable.
- **Optional:** `pip install youtube-transcript-api` turns on ad-read detection in video transcripts (best-effort, see Limits).
- ## Usage
- | Command | What it does |
- |---|---|
- | `/sponsor-sleuth @yourchannel` | Full scan: report, rate, 10 pitches, pipeline entries |
- | `/sponsor-sleuth pipeline` | Your pipeline, plus the follow-ups due today and their drafts |
- | `/sponsor-sleuth update NordVPN pitched` | Moves a deal along. `pitched` schedules follow-ups at day 4 and day 10 |
- | `/sponsor-sleuth update NordVPN closed 300` | Logs a closed deal. After 2, your rate calibrates from real prices |
- | `/sponsor-sleuth setup` | Sets the API key, currency, and your current rate (for the rate check) |
- Everything is written to `~/.sponsor-sleuth/runs/<handle>/<date>/`: `report.md`, `pitches/`, and every intermediate file, so you can audit how each brand got on the list.
- ## How it works
- ```
- profile ─▶ niche (Claude) ─▶ discover ─▶ prune (Claude) ─▶ scan ─▶ classify (Claude) ─▶ rank + reality check ─▶ contacts ─▶ price ─▶ report ─▶ pitches or growth plan (Claude)
- ```
- - **Scripts do the deterministic work:** YouTube API calls, caching, link decoding, regex, scoring, files. They live in `skills/sponsor-sleuth/scripts/` and use only the Python standard library.
- - **Claude does the judgment:** turning your channel into search queries, dropping off-niche lookalikes, classifying unclear signals (is this the creator's own merch or a sponsor?), and writing pitches. Claude is the host, so you don't need an Anthropic API key.
- **Detection layers**, strongest first:
- 1. The paid-promotion flag.
- 2. Multilingual disclosure phrases (EN/IT/ES/DE/FR/PT): "sponsored by", "sponsorizzato da", "patrocinado por", "gesponsert von", "#ad", "*Werbung"...
- 3. Promo codes next to brand links.
- 4. Affiliate and tracking links, decoded to the brand (Impact, Awin, CJ, ShareASale, Partnerize, LTK...).
- 5. Influencer UTM parameters.
- 6. Transcript ad reads.
- A link that appears in nearly every video of a creator is sent to Claude rather than counted, because it's usually their own shop or a gear list.
- **Each brand is labeled `sponsored`, `code` or `affiliate`:**
- - `sponsored`: paid evidence.
- - `code`: a promo code or tracked link with no disclosure.
- - `affiliate`: commission links only.
- The headlines count only `sponsored`. Affiliate-only brands get their own section, because they're often the easiest first yes.
- ### Quota
- The YouTube Data API's free tier is 10,000 units a day.
- | Step | Units |
- |---|---|
- | Profile | ~5 |
- | Discovery: up to 15 searches x 100 | ≤1,500 |
- | Featured channels + channel stats | ~25 |
- | Scanning 50 creators x 30 videos | ~100 |
- | Your own last 12 months | ~10 |
- | **Full scan** | **~1,650** |
- That's about 6 full scans a day. Searches are cached for a week and everything else for about a day, so reruns are close to free.
- ## Guardrails
- - **It drafts, you send.** No automated emailing or DMs. That keeps it GDPR-friendly and keeps your domain off spam lists.
- - **Every brand has a proof link.** Brands are never inferred from general knowledge.
- - **Public pages only.** It respects robots.txt and stays at 1 request per second per site. LinkedIn is never scraped; you get a search link instead.
- - **Your data stays local.** Your API key, runs and pipeline live in `~/.sponsor-sleuth/`. There's no telemetry.
- ## Limits
- - **YouTube only (v1).** Podcasts and Instagram/TikTok are on the roadmap.
- - **Transcripts are best-effort.** The official API only returns captions for your own videos, so transcript ad reads rely on `youtube-transcript-api`, which YouTube sometimes blocks. They're only used to name the brand when a video is flagged as paid but its description doesn't say who paid.
- - **Recall isn't 100%.** Undisclosed deals with no code or link can't be detected. The system is tuned so that what it does list is real.
- - **The rate is a model, not market data.** Nobody publishes what creators actually get paid. The CPM ranges in `references/cpm.json` are starting assumptions until you log your own closed deals.
- - **Status: v0.1.** It's fully tested offline against a simulated YouTube API (see `tests/`). Expect tuning after the first real-world runs.
- ## Development
- ```bash
- python -m unittest discover -s tests -v # unit + offline end-to-end, no key or network needed
- python skills/sponsor-sleuth/scripts/sleuth.py --help
- ```
- - **Add a phrase or a language:** edit `_SPONSOR_SRC` and `_CODE_SRC_CI` in `scripts/signals.py` and add a test in `tests/test_signals.py`.
- - **Add a domain** (an affiliate network, shortener, creator platform...): edit `references/link-domains.json`. No code change needed.
- - **Tune the ranking:** override any weight in `~/.sponsor-sleuth/config.json` under `"weights"` (see `DEFAULT_WEIGHTS` in `scripts/rank.py`).
- The original design notes are in [`docs/plan.md`](docs/plan.md).
- ## License
- MIT
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.