convez.ai
Skill
RevCentricGH

supersdr

RevCentricGH/supersdr

Ver no GitHub
InicialPluginScore · de 0 a 100

Claude skills for the SuperSDR community

Instalação

$npx convez-ai RevCentricGH/supersdr

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
1
Forks
0
Acompanhando
0
Saúde
42%
Último commit
há 1 mês
Versão
latest
Issues abertas
4
Contribuidores
3

README

A tradução para o português deste README ainda está na fila.

SuperSDR Skills

The ultimate toolkit of AI skills for running your AI-native cold-calling agency. Specifically curated for the SuperSDR community and supported by RevCentric.

Each skill teaches Claude to do one piece of the job such as: research a client, write a call script, build a dial-ready list, set up campaigns, follow up after a call, and track results.

What this is

This toolkit covers the two jobs that make up the business:

  1. Win a client. Land an agency customer: prep for the call, follow up after discovery, and send the proposal.
  2. Run outbound. Do the calling work for that client: set up Apollo, build the client brief, write the pitch, build the list, launch campaigns, and track performance.

Each skill is self-contained. You can run one on its own, or run several in sequence to cover a whole motion. Start with one skill and add more as you get comfortable.

How it works (the 60-second model)

Four things to know:

  • A skill is a packaged job. It is a folder of instructions (and sometimes helper files) that teaches Claude exactly how to do one task the SuperSDR way. You do not configure anything in code.
  • You run skills inside Claude. Most run in the Claude Cowork desktop app: download the skill's ZIP, upload the folder into a chat, and follow the prompts.
  • Two motions. The skills split into Win a client and Run outbound. Pick the motion you are in and work down the list. One utility sits outside both: grill-me, for planning any task before Claude runs it.
  • One doc ties outbound together: the client SPOT. A client SPOT is a client's Single Point of Truth, a multi-tab Google Doc with their positioning, ICP, and pain. You build it once per client with client-spot, and every other outbound skill reads from it.

Any jargon you hit along the way (ICP, TAM, disposition, sequence) is defined in the Glossary at the bottom.

Here is how the outbound skills connect once the SPOT exists:

apollo-account-setup   (run once per Apollo account, before any campaigns)

client-spot  ->  the SPOT doc
  |
  |-- Tab 3 + Tab 4  ->  cold-calling-screenplay
  |-- Tab 7          ->  objection-drill   (optional rep training, anytime)
  |-- Tab 9          ->  tam-contact-mapper  ->  list-builder  ->  apollo-campaign-builder
  |-- activated leads ->  custom-decks
  |-- dialer calls    ->  master-tracker

Run apollo-account-setup once per Apollo account. Then run client-spot first for each client. Everything else builds off the SPOT.

The autonomy ladder

Every skill sits on one of three rungs. The rung tells you what setup it needs before you start. As your comfort with AI tools grows, you climb the ladder.

  • Tier 1 · Cowork. You upload the skill into the Claude desktop app and run it in a Cowork chat or session. No terminal, no code, nothing to install beyond the app and a few connectors that are already available in Cowork. This is where everyone starts, and most of the toolkit lives here.
  • Tier 2 · Claude Code. The skill runs in a terminal as real software, with Python, local credentials, API connections. This rung is for jobs that need a runtime: pulling call data on a schedule, or rendering files like slide decks. You install it once and run a command. Three skills live here: master-tracker, custom-decks, and weekly-checkin.
  • Tier 3 · n8n + Claude Code. Fully automated pipelines where n8n triggers and orchestrates the work and Claude Code carries it out, with no human in the loop. This is the top rung and the direction the toolkit is heading. Nothing here yet, we are currently shipping it for RevCentric stay tuned :)

Level up your stack. Start at Tier 1 and get a feel for running skills in the Claude desktop app. When you want unattended reporting or branded decks, install the two Tier 2 skills. Tier 3 (n8n + Claude Code) is the horizon: the same skills, wired to run on their own. If you want a middle step toward automation, you can also push events into a live Cowork session from your phone or a webhook; see CHANNELS.md.

The tier is shown as a badge on every skill in the tables below.

The skills

Skills are grouped by the job you are doing. Each row shows the skill, what it does, its tier badge, and a one-click download.

Win a client

The acquisition motion: book the call, work the call, make the offer, kick off onboarding when they sign. The skills run in call order. post-discovery-followup hands off to client-proposal-doc-builder when the call ends in a proposal, and onboarding-kickoff starts the handoff to fulfillment once the client pays:

pre-brief  ->  discovery call  ->  post-discovery-followup
                                     |
                                     |-- proposal or follow-up outcome  ->  client-proposal-doc-builder

client signs and pays  ->  onboarding-kickoff  ->  client-spot (start fulfillment)

Stage 1 - Prep for the call

SkillWhat it doesTier
pre-briefTurn a booking-call transcript into a one-page meeting brief so you walk in prepared. Pulls the concerns, objections, asks, and commitments, each anchored to the moment in the call, and returns a Google Doc.Tier 1 · CoworkDownload ZIP

Stage 2 - After the call

SkillWhat it doesTier
post-discovery-followupRead a discovery-call transcript, decide the outcome, and take the next step. On a proposal or follow-up outcome it hands off to client-proposal-doc-builder for the draft, sends the approved email through Gmail, and updates the deal stage in Apollo. You approve every action.Tier 1 · CoworkDownload ZIP

Stage 3 - Make the offer

SkillWhat it doesTier
client-proposal-doc-builderBuild a send-ready outbound agency proposal from your discovery call as a Google Doc, with pricing tiers, the completed-conversations model, and T&Cs, then draft the follow-up email that sends the link.Tier 1 · CoworkDownload ZIP

Stage 4 - Kick off onboarding

SkillWhat it doesTier
onboarding-kickoffOnce a client signs and pays, draft and send the welcome and onboarding-form email, move the Apollo deal to the onboarding stage, and hand off to client-spot to start fulfillment. You approve the send and the stage move.Tier 1 · CoworkDownload ZIP

Run outbound

The fulfillment motion: the functional stages of running outbound for a signed client, in order.

Stage 1 - Set up Apollo (one time)

SkillWhat it doesTier
apollo-account-setupOne-time Apollo setup before any campaigns: link your workspace email, register your outbound number with the Free Caller Registry, and configure the 19 dispositions, 11 contact stages, and 19 disposition-to-stage triggers. Run once per Apollo account.Tier 1 · CoworkDownload ZIP

Stage 2 - Brief the client

SkillWhat it doesTier
client-spotBuild the client's SPOT: a multi-tab Google Doc covering campaign status, company overview, problem and solution, ICP, competitive landscape, objections, screenplay, and Apollo campaign setup. Every outbound skill reads from it.Tier 1 · CoworkDownload ZIP

Stage 3 - Build the pitch

SkillWhat it doesTier
cold-calling-screenplayWrite a word-for-word cold call script (Short or Full version) for any B2B company, built from the client's SPOT and live case studies pulled from their website.Tier 1 · CoworkDownload ZIP
objection-drillTrain reps on objections. Quick Drill gives three ready responses to a pasted objection; Live Roleplay plays the prospect, you respond, and it grades you. Covers the 5 core objection families.Tier 1 · CoworkDownload ZIP

Stage 4 - Build the list

SkillWhat it doesTier
tam-contact-mapperApply the client's ICP filters in Apollo's People tab and save the search as a named TAM view. Maps the full contact universe with no enrichment; nothing is imported. Run it before list-builder.Tier 1 · CoworkDownload ZIP
list-builderBuild an enriched, dial-ready contact list from the SPOT using Apollo. Scores fit and intent, routes each contact by temperature (Red Hot to Cold), and outputs a Google Sheet ready to dial.Tier 1 · CoworkDownload ZIP

Stage 5 - Launch the campaigns

SkillWhat it doesTier
apollo-campaign-builderSet up the client's full Apollo campaign: all 7 outreach sequences and 4 workflow plays, built in the Apollo UI for you. Run after list-builder.Tier 1 · CoworkDownload ZIP

Stage 6 - Advance the deal

SkillWhat it doesTier
custom-decksTerminal skill - runs in Claude Code, not Cowork (see Claude Code path below). Builds a tailored prospect deck from a call transcript and the prospect's website, branded as your agency, rendered to Google Slides and PDF with a View link.Tier 2 · Claude CodeDownload ZIP

Stage 7 - Track and measure

SkillWhat it doesTier
master-trackerTerminal skill - runs in Claude Code, not Cowork (see Claude Code path below). Pulls each rep's Apollo dialer calls into per-rep tabs of a Google Sheet, filtered to the dispositions you care about, deduped, and safe to re-run. Reads campaign health at a glance.Tier 2 · Claude CodeDownload ZIP
weekly-checkinTerminal skill - runs in Claude Code, not Cowork (see Claude Code path below). Builds a weekly per-client digest from the master-tracker sheet and each client's SmartLead stats, then delivers it to a configured destination (a Google Doc, Slack, or email). Runs on a weekly cron or on demand.Tier 2 · Claude CodeDownload ZIP

Work smarter

Cross-cutting utilities that improve every other skill and task.

SkillWhat it doesTier
grill-meA question loop for planning anything. Run it at the start of a session: it interviews you one question at a time until Claude fully understands the goal, inputs, outputs, and tools - then the work proceeds in the same session, fully briefed. A sharp plan beats a smart model.Tier 1 · CoworkDownload ZIP

Getting started

Start with the Cowork path. It is the simplest and covers every Tier 1 skill, which is most of the toolkit. The Claude Code path is for the two Tier 2 skills and is kept separate below; skip it until you naturally find gaps with Cowork and you genuinely feel the need to advance. You'll know when it happens, don't overengineer for a solution or pain you don't even feel yet.

Cowork path (start here)

Run a skill, in four steps:

  1. Pick a skill from the tables above and click its Download ZIP. You get a ZIP of just that skill folder.
  2. Extract the ZIP.
  3. Open the Claude desktop app, start a chat, and upload the entire extracted folder, not just SKILL.md. Some skills need their companion files.
  4. Follow the skill's prompts.

Whole-repo alternative: click the green Code button at the top of this page, choose Download ZIP, extract it, and upload the folder for the skill you want.

One-time Cowork setup

Do this once before running skills that reach external systems.

  • Install the Claude desktop app. Required for Cowork. Download it at https://claude.com/download.

  • Paste the voice rules into Project Instructions (recommended). Go to Claude, then Settings, then Project, then Instructions, and paste this so output sounds like a person across every session:

    Voice rules - apply to all output:
    - No AI-tell openers: "Great question", "Absolutely", "Certainly", "Of course"
    - No hedging: "I think", "it seems", "potentially", "it's worth noting"
    - No AI vocabulary: "delve", "leverage", "utilize", "robust", "seamless", "comprehensive", "cutting-edge"
    - No em-dashes. Hyphen or rewrite.
    - Short paragraphs. Write like a person, not a vendor.
    

    Each skill embeds these rules too; this just applies them globally.

  • Connect the connectors each skill needs. A connector links Claude to an outside service.

    • Google Drive: Settings, then Connectors, then Google Drive. Connect your account and enable edit access (read-only is not enough for the proposal builder or list-builder).
    • Gmail: Settings, then Connectors, then Gmail, with send access. Used by post-discovery-followup to send the approved follow-up and by onboarding-kickoff to send the welcome email, and only after you approve the recipient, subject, and body.
    • Apollo MCP: Settings, then MCP Servers, and connect Apollo (apollo-io). The skill calls Apollo through the connector, so no API key lives in the skill.
    • Browser automation (Claude in Chrome): Settings, then Computer Use, and enable browser control. Log into Apollo in Chrome and keep that tab open while running these skills.

Which connectors each skill needs:

SkillGoogle DriveApollo MCPBrowser automation (Claude in Chrome)GmailOther MCPs
apollo-account-setuprequired
client-spot
cold-calling-screenplay
tam-contact-mapperrequiredrequired
list-builderrequiredrequiredmany optional (see below)
apollo-campaign-builderrequired
objection-drill
client-proposal-doc-builderrequired (write)
post-discovery-followuprequiredrequiredrequired
onboarding-kickoffrequiredrequired
pre-briefrequired (write)

Optional MCPs for list-builder. Each one unlocks a stage of the pipeline if connected, and the skill degrades gracefully without it. Connect whichever you have.

MCPWhat it adds
PerplexityRicher Layer 4 signal research (otherwise falls back to native web search)
ZeroBounceReal email validation (otherwise relies on Apollo's verified flag)
TwilioPhone line-type detection (mobile / landline / VoIP)
ClayFallback contact enrichment when Apollo's hit rate is low
Common RoomIntent signals (website visitors, community engagement)
Smartlead / InstantlyPush the finished list to a cold email campaign
HeyReachPush the finished list to a LinkedIn outreach sequence
HubSpotSync contacts into CRM as leads

Advanced: Claude Code path

Skip this section if you only use the desktop app. It covers the three Tier 2 skills, master-tracker, custom-decks, and weekly-checkin, which run in a terminal with real Python, local API keys, and a Google OAuth token file. Do not upload them to Cowork. They cannot run there.

You drive these the same way you drive any skill, by asking Claude. The only difference is the home: instead of the desktop app, you run Claude Code in a terminal and tell it what you want in plain English ("set up master-tracker", "build a deck for this prospect"). Claude Code does the terminal work, so you should never have to type raw commands to run a skill.

Install your terminal and agent (one time).

  • Warp - the terminal we recommend. It is an AI terminal, so on the rare command Claude Code cannot run itself (an interactive login, a bit of debugging) Warp can help you directly. Download the desktop app at https://www.warp.dev/download and sign in.
  • Claude Code - the agent that runs the skills. Open Warp, install it with curl -fsSL https://claude.ai/install.sh | bash, then type claude to launch it and sign in. (Docs: https://code.claude.com/docs/en/setup.)

That install command is the only thing you type by hand. From here on you talk to Claude Code.

Credentials. These come from web dashboards, so you grab them yourself, then paste them to Claude Code and it writes them into the skill's config.json for you.

  • Apollo API key - required by master-tracker and custom-decks for call data and contact lookup. Apollo, then Settings, then Integrations, then API.
  • Google OAuth token - required by all three to work with Google Sheets and Drive. Each skill walks you through the OAuth setup and needs its own token.json. weekly-checkin reads the master-tracker sheet (read-only); Google Doc delivery additionally appends to a Doc.
  • SmartLead API key - required by weekly-checkin for per-campaign stats. SmartLead, then Settings, then API.
  • Deepgram API key - required by custom-decks for audio transcription. Sign up at deepgram.com.
  • Groq API key - required by custom-decks as a transcription fallback; fires only when Deepgram returns empty. Sign up at console.groq.com.

Run it. With Claude Code open in the skill's folder, ask it to "set up and run master-tracker" or "build a custom deck for this prospect". It installs the dependencies, fills in config.json from the credentials you pasted, and runs the skill.

Prefer to run the commands yourself? Here is what Claude Code does under the hood:

# master-tracker
cd master-tracker
pip install -r requirements.txt
cp config.template.json config.json   # fill in your credentials
python3 run.py

# custom-decks (also needs Node and Marp for deck rendering)
cd custom-decks
pip install -r requirements.txt
npm install -g @marp-team/marp-cli
cp config.template.json config.json   # fill in your credentials
python3 run.py

# weekly-checkin (needs a SmartLead API key and the master-tracker Google Sheet)
cd weekly-checkin
pip install -r requirements.txt
cp config.template.json config.json   # fill in clients + delivery destination
python3 run.py --week 2026-W22 --dry-run   # preview, then drop --dry-run to deliver

Recommended path by situation

  • Brand-new agency, no clients yet. Run the Run outbound pipeline on yourself to book your first meetings, with your own agency as the client: apollo-account-setup, then client-spot for your own offer, then down the stages. Use Win a client to work the calls you book.
  • Just signed a client. Move to Run outbound and work the stages in order. Run apollo-account-setup once (only the first time on a new Apollo account), then client-spot to build the SPOT, then down the list: pitch, list, campaigns.
  • Already running campaigns. Install master-tracker (Tier 2) to pull call data into a sheet and read campaign health. Use custom-decks (Tier 2) to build decks for prospects who are warming up.
  • Debugging a campaign that is not working. Re-read the SPOT first; weak positioning shows up everywhere downstream. Re-run list-builder on a small sample to sanity-check the ICP before scaling. Check master-tracker for conversion rates by rep. If reps stall on calls, run objection-drill.
  • Starting a big or unfamiliar task. Run grill-me first. It interviews you until the goal, inputs, and outputs are airtight, then the work proceeds in the same session - fully briefed.

Glossary

Plain-language definitions of the jargon you will hit.

  • Skill - a packaged job for Claude: a folder with instructions and sometimes helper files. You run it; you do not edit it.
  • Cowork vs Claude Code - two ways to run a skill. Cowork is the Claude desktop app, where you upload a skill and run it in a chat (Tier 1). Claude Code is the terminal, where a skill runs as real software (Tier 2).
  • SPOT - a client's Single Point of Truth: the multi-tab Google Doc holding their positioning, ICP, and pain. Built once per client by client-spot; the outbound skills read from it.
  • ICP - Ideal Customer Profile: the type of company and person a client wants to reach (industry, size, title, signals). It drives who lands on the dial list.
  • TAM - Total Addressable Market: the full universe of contacts that fit the ICP. tam-contact-mapper saves this as an Apollo search before any enrichment.
  • Dialer - the software reps use to make calls. In this toolkit, reps dial in Apollo's dialer, which logs each call.
  • Disposition - the outcome you tag on a call (for example "Meeting Scheduled" or "No Answer / Not Available"). apollo-account-setup configures the standard set, and master-tracker filters and counts by disposition.
  • MCP connector - a link that lets Claude talk to an outside service (Apollo, Google Drive, Gmail) without you handing it an API key inside a skill. You connect it once in settings.
  • Browser automation (Claude in Chrome) - Claude controlling a real Chrome tab to click through a web app, used for the Apollo UI work where there is no connector.
  • Sequence and workflow play - Apollo outreach building blocks. A sequence is a series of steps (calls, emails) a contact moves through; a workflow play automates an action when a condition is met. apollo-campaign-builder sets these up.

Cost

What you pay for, plain:

  • A paid Claude plan (required). Cowork runs in the Claude desktop app, which needs a paid Claude subscription. The two Tier 2 skills also use your Claude access from the terminal. This is the one cost everyone has. Start with the pro plan, if you start maxing out your usage quickly upgrade to Max.
  • An Apollo plan (required for outbound). Apollo provides the database, contact data, and the MCP connector. A paid Apollo plan is what the outbound skills run on.
  • Optional add-ons (only if you connect them). The Tier 2 custom-decks skill uses Deepgram and Groq for transcription, both with free tiers and pay-as-you-go pricing. The list-builder optional MCPs (ZeroBounce, Twilio, Clay, Perplexity, and the rest) are each separate accounts you only pay for if you turn them on. Everything optional degrades gracefully when it is off, so you can start with none of them.

You do not pay this project anything. The skills are free and open. Your only costs are the third-party accounts above.

Compliance

Outbound calling is regulated. You are responsible for running it legally. A few things to get right before you dial:

  • Call-recording consent. Some jurisdictions require every party on a call to consent before you record. Know the rules where you and the prospect are, and get consent when it is required.
  • Do Not Call. Honor Do Not Call lists and any opt-out a prospect gives you. list-builder's dial-ready gate excludes do-not-call and already-pulled contacts, but the responsibility to scrub and honor DNC and suppression is yours.
  • This is not legal advice. Rules vary by country, state, and industry. When in doubt, check with counsel before launching a campaign.

Stay updated, contributing, and license

Stay updated. Skills improve over time. Watch this repo: click Watch, then Custom, then Releases only at the top of this page. GitHub emails you whenever a new version ships. When a release lands, re-download and re-upload the affected skill's ZIP. All releases live at https://github.com/RevCentricGH/supersdr/releases.

Contributing. Found a bug or have an improvement? Open a GitHub issue, or send a pull request. Note that everything in this repo ships publicly as per-skill ZIPs, so do not commit anything private.

License. These skills are published openly for the SuperSDR community. Anything committed here is public. No formal license is attached yet; if you plan to reuse the skills outside the community, open an issue to ask.

Ver o original no GitHub

Relacionados