The Digital Foundry
Find my order
AgentTDF-AG-03
Lead Research Agent
Wake up to a list you'd actually want to call.
THE DIGITAL FOUNDRY$149

Agent · 03

Lead Research Agent

An agent that researches prospects against criteria you define, checks each one against basic fit signals, scores it, and writes a two-line brief per lead — who they are, why they qualify, one opener. Run it inside Claude in a browser tab with nothing to install, or deploy the included app and it runs unattended from a CSV using Claude's built-in web search — no third-party search key. Works with Claude — Track A needs only a free account to try.

$149one-time · yours forever
Instant delivery — download page opens right after checkout

All sales final — files arrive instantly, so there’s nothing to return. Broken or not as described? We fix it: caremindcontactus@gmail.com

What’s inside

Open the folder before you buy it.

This is the exact manifest of the download — no mystery box. A full excerpt is free below, so you can judge the quality before paying.

Read the free sample →
lead-research-agent/ · 40 files · 132 KB
START-HERE.mdplain text
icp-template.mdplain text — you edit this
system-prompt.mdplain text
scoring-rubric.mdplain text
agent.config.json·
claude-project/4 files — Track A
SETUP.md · project-instructions.md · kickoff-prompt.md · troubleshooting.md·
app/16 files — Track B
README.md · .env.example · sample-icp.md · sample-leads.csv · package.json·
src/ (6 files) · test/ (5 files)·
prompts/3 files
tools/4 files
search-prospects.json · check-fit.json · write-brief.json · dedupe.json·
templates/lead-brief-template.md1 file
examples/2 files
eval-cases/leads-30.jsonl30 cases
deploy-readme.mdplain text

Free sample · scoring-rubric.md

The no-evidence rule — silence is not a mid-range score.

A category with no supporting evidence scores 0 for that category. Do not average toward the middle when information is missing — missing information is itself information (the signal is weak or the source didn't have it), and the score should reflect that honestly.

Score drift toward the middle is the main failure mode of any scoring agent — it creeps as the model tries to be "fair" to ambiguous cases. That's why eval-cases/leads-30.jsonl exists: run it any time you change the rubric or the ICP, and if scores drift more than about 10 points from expected, tighten the rubric before trusting it on real prospects again.

— excerpt ends. Full four-category rubric and score bands in the complete package.

Buy it if. Skip it if.

Built for you if

  • You have clear criteria for a good-fit lead and just need the list built
  • You want it running today, not a build project — Track B ships with a sample ICP and sample leads so you see output before touching your own data
  • You're fine spending fifteen minutes filling in your own ICP — vague criteria produce vague scores

Skip it if

  • You don't yet know who your ideal customer is — this agent qualifies against criteria, it doesn't define them
  • You want a hosted service with a dashboard — this runs inside your own Claude account or your own machine
  • You don't use Claude or want an Anthropic API key — Track A needs a free Claude account, Track B needs an API key

Questions