agentleFS
BenPremBenPrem/business-brain/.claude/skills/cold-email-campaign/SKILL.md

cold-email-campaign

Skill1 starsChanged 3 months agoMITView on GitHub

Batch cold-email pipeline — load new-lead records from the workspace CRM, generate personalized first-touch emails, export a CSV ready for <YOUR OUTREACH TOOL>. Trigger on "batch", "campaign", "run the campaign tool", or first-touch emails for 5+ leads at once. All writing rules live in cold-email. NOT for leads already contacted or in-pipeline (use follow-up-nurture) or a single hand-written email (use cold-email).

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 6 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    No percentages or statistics you can't verify — never invent data

    line 28
  2. Distinctive:

    No differentiator / "what sets us apart" pitch — that belongs in Touch 3

    line 30
  3. Distinctive:

    No asking for a 30-minute call on first touch — the CTA is a small yes

    line 31
  4. Distinctive:

    Ask the founder if not provided: niche (industry filter, blank = all), limit (default 20).

    line 40
  5. Distinctive:

    Import the CSV as a new sequence.

    line 73
  6. Very specific:

    Update the workspace deal record to stage: contacted and log an activity note (type: email-sent)

    line 83
  7. Distinctive:

    Use a cheap model tier for batch generation and validate the model slug against your router's live model list before any run — slugs rot.

    line 91

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Cold Email Campaign Builder

Batch workflow only: pull leads → generate copy at scale → export a CSV for <YOUR OUTREACH TOOL>. All writing rules, tone, structure, and subject-line craft live in the cold-email skill — this skill is the assembly line, not the copywriter.

The batch script is tools/cold_email.py. It reads lead records, calls your LLM router with a system prompt encoding the cold-email rules, and writes the export CSV. Everything below also works manually (agent-generated, one lead at a time) if you haven't wired the script's API key yet.

Where leads come from

tools/cold_email.py reads deal records from your workspace pipeline folder with stage: new (accepts "New Lead" as an alias). Each record needs frontmatter name, industry, location, website, and a Pain points: ... line in the notes or body — these are what personalization is built from; a record without pain points produces a generic email, which defeats the point. lead-scraper creates these records; verify they exist on disk before running a campaign — a scored list that never got written is not a pipeline.

What NOT to include in Touch 1

  • No percentages or statistics you can't verify — never invent data
  • No "we helped X company increase Y by Z%" — case-study proof belongs in Touch 2
  • No differentiator / "what sets us apart" pitch — that belongs in Touch 3
  • No asking for a 30-minute call on first touch — the CTA is a small yes

Target length

Under 75 words. Hard cap — it wins over any other number found anywhere.

Batch workflow

Step 1 — Confirm parameters

Ask the founder if not provided: niche (industry filter, blank = all), limit (default 20). Always recommend a dry run for a new niche.

Step 2 — Verify leads load (free, no API calls)

python3 tools/cold_email.py --list --niche "NICHE"

Zero leads → create or fix the pipeline records first; don't proceed to generation.

Step 3 — Dry run (~5 emails, real API calls)

python3 tools/cold_email.py --stage "New Lead" --niche "NICHE" --limit 5 --dry-run

The founder reads the previews and confirms tone before the full batch burns budget on 20 emails in the wrong voice.

Step 4 — Full run

python3 tools/cold_email.py --stage "New Lead" --niche "NICHE" --limit LIMIT

The CSV lands in ventures/<your-venture>/deliverables/cold-email-exports/ (created at runtime).

Step 5 — Report results

How many generated, preview of the top 3 (subject + body), CSV path.

Before sending — email addresses

Lead records usually capture name/website/phone, not email addresses. Fill first_name, last_name, email in the CSV via the business website, an email-finder tool's free tier, or a phone call. Never send to a guessed or unverified address.

Importing into <YOUR OUTREACH TOOL>

  • Import the CSV as a new sequence; default cadence: 3 touches, 3 days apart
  • Send from a secondary domain — never your primary business domain (deliverability damage on your main domain is near-permanent)
  • Subject line is pre-written in the CSV — use as-is

After the founder confirms send (non-negotiable)

For each sent lead:

  • Move its <TASK SYSTEM> task to the Contacted stage and comment "Cold email touch 1 sent YYYY-MM-DD"
  • Update the workspace deal record to stage: contacted and log an activity note (type: email-sent)

The script does NOT do these writes — the agent does, then verifies them on disk. When replies come in, hand off to follow-up-nurture.

Costs and hygiene

  • Use a cheap model tier for batch generation and validate the model slug against your router's live model list before any run — slugs rot.
  • A 20-email batch costs on the order of cents at budget-tier pricing; still dry-run first.
  • The export CSV contains prospect contact data — gitignore the exports folder; never commit it to a repo.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.