agentleFS
automatewithudayautomatewithuday/GTM-Public-Traded-AI-Accounts/.claude/skills/icp/SKILL.md

icp

Skill0 starsChanged 4 months agoMITView on GitHub

Interactively define a custom ICP (Ideal Customer Profile) and write it as a profile pack under profiles/<name>/ — sector vocabulary, scoring weights, tier cutlines, services, and voice. Use when a user wants to run AIPT for a different business or vertical than the default martechs.io pack, or wants to build or edit a profile pack.

3 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions3 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 3

    1. echris6/ai-business-skills★ 16same name
    2. kasperjunge/skills★ 7same name
    3. marek-kujda/ad-copy-skill-pack★ 4same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository8 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 2 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Ask for 3-5 of the user's best customers (bought, renewed, or expanded) and 2-3 worst-fit ones (churned, never closed, wrong fit).

    line 16
  2. Distinctive:

    Ask for the qualify/disqualify thresholds, or infer them by asking where the user's best/worst customers would land on the point scale from …

    line 38
  3. Very specific:

    Ask what verticals this ICP targets.

    line 45
  4. Very specific:

    Ask what services/offers this business sells → services.yaml, same shape as config/services.yaml (per service: name, description, …

    line 55
  5. Very specific:

    Ask for the pack <name> if not given, write the files to profiles/<name>/, and optionally capture the interview's raw ICP narrative in …

    line 64

≈ “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

/icp — build a profile pack (6-step interview)

Pure conversation — no API calls, no subagents, no cost. Ends by writing profiles/<name>/settings.yaml (and only the other files that differ from the default pack: services.yaml, personas.yaml, outbound_copywriter.md, icp.md) and validating the result. Profile packs are directory overlays — any file missing from the pack falls back to config/.

Step 1 — best/worst customers

Ask for 3-5 of the user's best customers (bought, renewed, or expanded) and 2-3 worst-fit ones (churned, never closed, wrong fit). Capture what each side has in common — this contrast drives everything that follows.

Step 2 — discriminating attributes

From the best/worst contrast, extract 3-6 attributes that actually separate good-fit from bad-fit (e.g. "recent leadership change", "raised money in the last 12mo", "no in-house AI/eng team") — not generic firmographics. Map each onto the signal taxonomy shape in docs/SIGNALS.md (E1-E9 / P1-P6) where it fits; flag genuinely new attributes as candidate signal types.

Step 3 — point values

For each attribute, ask how strongly it should count relative to the others. Translate into scoring.weights entries and scoring.component_caps (intent / capability_gap / timing / commercial_fit), following the structure in config/settings.yaml. Caps conventionally sum to 100 — the validator warns (non-fatally) if they don't.

Step 4 — tier cutlines

Ask for the qualify/disqualify thresholds, or infer them by asking where the user's best/worst customers would land on the point scale from step 3. Write scoring.qualify_threshold and scoring.disqualify_below (the gap between them is the human-review band).

Step 5 — sector/SIC vocabulary

Ask what verticals this ICP targets. Sector vocabulary is free text — any lowercase key works; it does not need to match the built-in saas/fintech/edtech/healthcare set. For each sector, collect SIC codes (if known) and/or name/description keywords, matching the universe.sectors.<name>.sic / .keywords / .exclude_sic shape in config/settings.yaml. Also confirm universe.exchanges and the market_cap_min/market_cap_max band.

Step 6 — services + voice

Ask what services/offers this business sells → services.yaml, same shape as config/services.yaml (per service: name, description, ideal_signals, pitch_angle). Then ask about voice/tone for outreach copy — banned words, CTA style, sign-off — → outbound_copywriter.md, modeled on config/outbound_copywriter.md. Skip either file if the default pack's version already fits (overlay fallback covers it).

Write + validate

Ask for the pack <name> if not given, write the files to profiles/<name>/, and optionally capture the interview's raw ICP narrative in profiles/<name>/icp.md for future reference. Then validate:

uv run python -m pipeline --profile <name> profile --validate

Report errors back to the user and iterate — never leave an invalid pack on disk silently. Close by showing how to use it: uv run python -m pipeline --profile <name> discover --dry-run (or AIPT_PROFILE=<name>).

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.