Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/cmj-hub/claude-cold-emailnpx agentmods add skills/cmj-hub/claude-cold-email/cold-email-reply-scoringWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/cmj-hub/claude-cold-email/cold-email-reply-scoring)<a href="https://agentmods.dev/skills/cmj-hub/claude-cold-email/cold-email-reply-scoring"><img src="https://agentmods.dev/badge/skills/cmj-hub/claude-cold-email/cold-email-reply-scoring/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/cmj-hub/claude-cold-email/cold-email-reply-scoring"><img src="https://agentmods.dev/badge/skills/cmj-hub/claude-cold-email/cold-email-reply-scoring.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00094 | $0.01566 |
| Opus 5 | $0.00047 | $0.00783 |
| Sonnet 5 | $0.00019 | $0.00313 |
| Haiku 4.5 | $0.00009 | $0.00157 |
Grade A, and why
cold-email-reply-scoring scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 174 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Cold Email Reply Scoring — deterministic reply classifier
Real classifier, not vibes. Backed by scripts/score_reply.py for
deterministic categorization. The output is a category + confidence
- recommended routing action.
Activation
Loaded by:
cold-email-weekly-rhythm— Wednesday + Friday reply triage- User invocation: "Score this reply", "Classify this response"
Output categories
The classifier returns one of 4 categories with a confidence score:
| Category | Signal | Routing (per brand-config) |
|---|---|---|
buy-signal |
Explicit ask to talk, calendar request, price question, decision-maker tag | brand-config.operations.reply_routing.buy_signal |
positive |
Curious, asking follow-up Qs, expressing interest, asking for proof | positive |
neutral |
"Not now", "circle back later", "send more info" without commitment | neutral (often → nurture stream) |
not-interested |
Explicit no, unsubscribe ask, hostile, out-of-office on repeat | not-interested |
Feature set (what the classifier uses)
Pure feature engineering — no LLM, no vendor APIs:
-
Intent keywords — regex over 50+ patterns per category
- buy-signal: "send the calendar", "what's the price", "demo", "case study", "intro"
- positive: "tell me more", "interesting", "curious", "send the deck"
- neutral: "not the right time", "Q4", "send more info", "in a few months"
- not-interested: "unsubscribe", "stop emailing", "not a fit", "no thanks"
-
Time-to-reply — minutes since send
- <15 min after send → likely auto-reply (out-of-office, etc.)
- <60 min → high-intent (operator was working, saw it, replied immediately)
- 60 min - 24h → engaged reply
-
24h → background reply
-
Reply length — chars
- <20 chars → 1-word answer, usually not-interested or short positive
- 20-200 → typical engaged reply
-
200 → often a substantive response (rare; high-signal)
-
Question count — # of "?" in the body
- 0 → declarative (could be any category)
- 1-2 → engaged with specific questions
- 3+ → either highly engaged or asking-for-everything (lower convert)
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago Changed · -1 lines 92211bfd7491
- 10d ago First seen · 175 lines · 94 tokens per session scan A 72c1a45041e5
cold-email-reply-scoring is a skill published in the GitHub repository cmj-hub/claude-cold-email (2 stars, last pushed 2d ago), licensed MIT. It adds 94 tokens to every session and 1,566 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
afrexai-lead-hunter
Enterprise-grade B2B lead generation, enrichment, scoring, and outreach sequencing for AI agents. Find ideal prospects, enrich with verified data, score against your ICP, and generate personalized outreach — all autonomously.
graphify
Knowledge graph engine for B2B sales intelligence. Builds queryable graphs from product catalogs, customer conversations, and market research. Powered by graphify.
telegram-toolkit
Register these commands with @BotFather using /setcommands.
lead-discovery
AI-driven lead discovery for B2B export. Searches web for potential buyers matching ICP, evaluates fit, and creates CRM records for follow-up.
b2b-sdr-agent
Open-source B2B AI SDR template. 7-layer context system with 10-stage sales pipeline, 4-layer anti-amnesia memory, 14 automated pipeline checks, WhatsApp IP isolation, multi-channel (WhatsApp+Telegram+Email). Built on OpenClaw.
sdr-humanizer
Transform AI-generated sales messages into natural, human-like conversations that build trust and rapport.