bama
BAMA: 모든 Mac 앱을 만들 수 있습니다
Your agent forgets the brief every new thread마다 같은 동작을 그대로 실행하고 실패를 살펴봅니다
Your agent forgets the brief every new thread. The same moves, run by your agents, inspectable when they fail.
스킬JavaScript · PythonGitHub @cmj-hub온라인운영자 미확인
써 보셨어요? 한 번만 눌러 주세요. 로그인은 필요 없어요.
개인 계정의 공개 저장소 소유자는 GitHub로 로그인해 운영자를 확인하고 소개를 직접 관리할 수 있습니다. 관리자 승인은 필요하지 않습니다.
로그인한 뒤 확인 파일을 만들어 이 저장소의 기본 브랜치에 30분 안에 커밋해 주세요. 다른 브랜치나 포크의 파일은 인정하지 않습니다. 파일을 올린 뒤 아래에서 확인하면 관리자 승인 없이 관리할 수 있습니다. 자격을 유지하려면 파일을 남겨 두세요.
로그인이 필요하면 로그인 화면으로 이동합니다. 로그인 후 이곳으로 돌아와 원하는 확인 버튼을 다시 눌러 주세요.
프로젝트 폴더에서 /nomorevibe verify를 실행하면 소유권을 확인하고 소개를 직접 관리할 수 있습니다.
1분 만에 확인하기Sales prospecting skill for Claude Code. Sales prospecting builds the B2B prospect list you are willing to write to. The scorer refuses a title-only list.
저장소 README에서 가져옴
README 전체를 불러오는 중입니다. 아래는 저장된 발췌입니다.
Sales prospecting builds the B2B prospect list you are willing to write to.
Ops lead at Northwind posted a role for an outbound lead this week.
The good draft says call this week. A title-only list fails the score.
The build guide teaches a human. The pack teaches an agent.
The skill then loads as prospect-list:who-to-contact. Ask Claude who to call this week, or hand it a list of titles.
For every agent host the skills installer knows:README 전문 보기 ↗Sales prospecting builds the B2B prospect list you are willing to write to.
/plugin marketplace add cmj-hub/gtm-operator-skills
/plugin install prospect-list@gtm-operator-skills
/prospect-list:who-to-contact
Or score the sample without an agent:
python3 skills/who-to-contact/scripts/score.py --file skills/who-to-contact/examples/list-good.json # exit 0, prints title, signal, score, then "Next: /cold-email:cold-email"
python3 skills/who-to-contact/scripts/score.py --file skills/who-to-contact/examples/list-title.json # exit 1: - missing signal (a title with no signal is a directory, not a list) → write one thing they did this window, or leave them off
Part of the GTM operator suite: /plugin install gtm@gtm-operator-skills installs all ten.
Add the gtm-operator mod to see the suite's next step above your prompt and keep brand-config.json from being overwritten: /plugin install gtm-operator@gtm-operator-skills.
Ops lead at Northwind posted a role for an outbound lead this week.
The good draft says call this week. A title-only list fails the score.
The build guide teaches a human. The pack teaches an agent.
As a Claude Code plugin:
/plugin marketplace add cmj-hub/claude-prospect-list
/plugin install prospect-list@claude-prospect-list
The skill then loads as /prospect-list:who-to-contact. Ask Claude who to call this week, or hand it a list of titles. /prospect-list:who-to-contact score scores the draft already in gtm/list.json.
For every agent host the skills installer knows:
npx skills add cmj-hub/claude-prospect-list --all -g --full-depth
One host:
npx skills add cmj-hub/claude-prospect-list --skill '*' -g --full-depth -y -a claude-code
Swap claude-code for cursor, codex, grok, github-copilot, windsurf, cline, or opencode. The scorer is Python in this repo. It does not call a paid API, and it does not buy data.
The same pack is on npm as @jmc-gtm/claude-prospect-list, released from this repo's version tags with npm provenance. It has no dependencies and no install scripts.
In a project, for any agent host the skills installer knows (experimental_sync is still marked experimental there):
npm install -D @jmc-gtm/claude-prospect-list
npx skills experimental_sync -a claude-code -y
In your own Claude Code plugin marketplace, point an entry at the package:
{ "name": "prospect-list", "source": { "source": "npm", "package": "@jmc-gtm/claude-prospect-list" } }
cd skills/who-to-contact
python3 scripts/score.py --file examples/list-good.json
python3 scripts/score.py --file examples/list-title.json
The good draft exits 0 and prints a list score from a signal. The title-only draft exits 1 and prints one - what is wrong → what to change line for each gap, then Next: fix the lines above and run this again. A persona label in place of a signal, or a score that is not call this week, hold, or drop, fails too. Add --json for one result object (fixes and next included). Then drop in yours at gtm/list.json in your project.
| Path | What it is |
|---|---|
skills/who-to-contact/SKILL.md | The skill: what a signal is, how to pick the slot, the checklist. |
skills/who-to-contact/scripts/score.py | The scorer. Exit 0 scored, 1 refused, 2 unreadable input. |
skills/who-to-contact/examples/ | Good, hold, title-only, and persona-label drafts. |
.claude-plugin/ | Plugin and marketplace manifests for Claude Code, and the plugin icon. |
tests/ | python3 -m unittest discover -s tests |
It will not send the list. It does not buy data. It will not accept a title-only list.
This pack scores who earns a slot this week from a public signal. Building the company universe and verifying emails sit outside the pack.
A found address is not sendable until verify returns deliverable.
The people a public signal puts on this week's list. A job title with no signal is not a prospect yet.
No. You bring the signal. The pack scores the slot: call, hold, or drop.
Growth Audit: architecture gaps in the GTM you already run. Free written report.
Friday Signal: one Friday GTM read. No pitch in it.
The scorer is standard-library Python and opens no network connection. It reads only the draft you give it; the skill writes gtm/list.json, reads brand-config.json and never writes to it. No telemetry, no credentials, no lookups, and nothing is sent. See SECURITY.md.
Previous: Email sequence
Next: GTM skills for Claude Code
MIT. Python 3 standard library only.