🔬 Research Report

Sales Training Repo Deep Dive

Compiled July 12, 2026 · Follow-up to the Landscape Research · All five recommended repos shallow-cloned, skills inventoried, training-relevant SKILL.md files read in full

← Back to Hub
📌

The short version

Summary

The decisive finding: no repo contains a role-play engine, a fear ladder, an audible loop, a certification gate, or a longitudinal scorecard. The ecosystem is 100% knowledge and execution, 0% practice. Your program's core is genuinely novel; the repos supply the curriculum content underneath it.

What that means in practice: install the louisblythe Core 20 as reference knowledge, mine zubair's massive objections file into the /objection-gym corpus, adopt the weighted scoring rubric for /call-review and /scorecard, copy northrn's engineering discipline as house style, and build /roleplay first, from scratch, because it's your stated top want and the one thing nobody provides.

Repos examined: louisblythe/Sales-Skills, coreyhaines31/marketingskills, zubair-trabzada/ai-sales-team-claude, sales-skills/sales, northrn/claude-sales-skills.

📚

1. louisblythe/Sales-Skills: 122 skills (THE knowledge base)

Install Core 20

Fork of coreyhaines31's format by Apparate. 122 SKILL.md files, roughly 7-18K characters each, consistently structured: frontmatter description → Initial Assessment → Core Principles → named framework → stage-specific guidance → Common Mistakes → Practice Exercises → "Questions to Ask" → Related Skills.

Quality is high and human-teaching oriented for the Core 20. A large middle tier (about 60 skills) is written for building sales bots (conversation-quality-scoring, dynamic-script-generation, emotional-arc-management, human-in-the-loop-training) with pseudo-Python scoring functions. The rest is ops and automation filler (timezone-awareness, voicemail-drop-optimization).

Skills worth stealing (read in full)

SkillWhat it containsSteal for
objection-handlingLAER framework (Listen, Acknowledge, Explore, Respond), objection taxonomy (price, timing, trust, need, authority), hidden-objection surfacing, what NOT to do, 5 practice drills including the "3-Response Drill" and "Pause Practice" (wait 3 seconds before responding)/objection-gym core curriculum
asking-effective-questionsFull SPIN + BANT + MEDDIC question banks, questions by stage, "Magic Wand" and "Consequence" priority-revealing questions, 5 questioning mistakes/roleplay discovery scenarios, /certify rubric
discoveryComplete discovery-call structure (opening → situation → problem → impact → future state → process → close), quantifying pain, discovery by motion (SMB vs mid-market fits Annette)/roleplay call scripts, /call-review checklist
closingBuying-signal recognition, 7 closing techniques, 6-step modern closing process, post-ask branches (yes / need time / no / dark), "Silence Game" and "Direct Ask Practice" drills/roleplay closing reps, /daily-rep
active-listeningHEAR framework, paraphrase / reflect / summarize techniques, 3-Second Rule drill, question-ratio tracking metric/call-review scoring dimension
resilienceRejection-as-data reframes, recovery routine, daily reset, energy management, warning signs/fear-ladder mindset layer (closest thing to fear work in ANY repo)
sales-psychology70+ mental models (loss aversion, status quo bias, anchoring, social proof...) organized by sales stage, with explicit ethical guidelinesPersona realism in /roleplay; teach-the-why moments
pricing-negotiationPricing objection types with responses, buyer tactics to recognize (flinch, deadline pressure), trade-don't-cave, walk-away criteria/roleplay pricing scenarios (her scariest moment)
building-rapportFirst 60 seconds, personality-type adaptation (analytical, driver, expressive, amiable), rapport-break recoveryPersona variety in /roleplay
conversation-quality-scoring5 weighted dimensions: engagement, discovery quality (pain 25, budget 20, timeline 20, process 15, stakeholders 15, current solution 10), value communication, objection handling, progression; benchmark plus coaching-recommendation logic/scorecard and /call-review scoring rubric, directly adaptable
deal-review-win-lossWin/loss question sets, loss-reason taxonomy, review meeting structure/call-review debrief format
emotional-arc-managementCuriosity → pain awareness → desire → confidence phase model with transition signals/roleplay feedback ("where did the arc stall?")
dynamic-script-generationComponent-library approach to assembling personalized scripts/pitch-polish variant generation
human-in-the-loop-trainingFeedback-loop design, correction processing, pattern learning/scorecard longitudinal improvement tracking

Verdict

Install-as-knowledge for the Core 20 is confirmed. Nothing here is a role-play engine: every skill teaches a human or configures a bot, none plays a prospect. The practice exercises across skills are a ready-made drill bank for /daily-rep.

🧲

2. coreyhaines31/marketingskills: 47 skills

Steal 2 patterns

Same family as the offers and positioning-basics skills you already run. One deep read: sales-enablement (14.5K chars) covers the 10-12 slide pitch deck framework, one-pagers, objection handling docs in two formats (quick-reference matrix plus deep-dive per objection with an Acknowledge → Reframe → Evidence → Question structure), ROI calculators, demo scripts and talk tracks, buyer persona cards, and sales playbooks.

Core principle worth keeping: "Sales uses what sales trusts: situation-specific, not generic."

Verdict

Steal the objection-doc two-format pattern for /objection-gym output (drill live, then generate her personal quick-reference card), and persona cards as the template for /roleplay personas. The rest is collateral production, already covered by her existing skills.

⚙️

3. zubair-trabzada/ai-sales-team-claude: 14 skills + 5 agents

Mine 2 now, hold rest

Command-style skills (a /sales prospect <url> orchestrator routes to sub-skills). Big, prescriptive, output-template-driven. Deep-read highlights:

  • sales-objections (40.7K chars, the largest single objection resource found): Feel-Felt-Found and Acknowledge-Bridge-Close frameworks, 15 universal objections each with word-for-word responses, industry-specific objection generation rules, competitive objections, a 5-tactic pricing deep dive (ROI reframe, cost of inaction, TCO, tier-down, payment terms), objection prevention, and a quick-reference matrix output. This is the seed corpus for /objection-gym.
  • sales-qualify (22.9K): BANT scored 0-25 per letter plus a MEDDIC completeness score, rolled into a 0-100 opportunity grade with lead-grade assignment. The scoring-with-evidence pattern is exactly the shape /certify and /scorecard need.
  • sales-prep (16.3K): meeting brief with a cheat-sheet card, 10 discovery questions, 5 objections to expect, competitive landmines. Pattern for pre-roleplay briefing docs tied to her real prospects (do-it-first demos, the Boulder GBP list).
  • sales-followup (20.1K): 5 scenarios (post-meeting, post-demo, post-proposal, ghost recovery, nurture) with full email copy, voicemail scripts, and a cadence matrix.

Verdict

Hold for the execution phase as the landscape doc said, EXCEPT sales-objections (mine now for /objection-gym drill content) and the qualify scoring pattern (mine now for /scorecard).

🗄️

4. sales-skills/sales: 500+ skill folders

Confirmed skip

2,349 files, almost entirely per-platform tool integrations (sales-6sense, sales-activecampaign, sales-adroll, sales-ahrefs, ...), each with SKILL.md plus api-reference plus platform-guide. Zero training content. Nothing to adapt.

🛠️

5. northrn/claude-sales-skills: 7 skills, small but well-engineered

Copy the discipline

Not SKILL.md format (plain .md plus a README per skill) but disciplined: explicit Tool Restriction and Token Isolation sections, step-numbered workflows, dynamic tuning, model-routing notes. Skills: account-research, competitor-intel, discovery-questions, email-writer, lead-generation, meeting-prep, proposal-writer.

  • discovery-questions: generates role- and industry-tailored SPIN question sets with follow-up prompts, red flags, and a timed 30-minute call structure (opening 3 / situation 5 / problem 10 / implication 5 / qualification 5 / next steps 2). That timed structure is a ready /roleplay session skeleton.
  • meeting-prep: a compact version of zubair's prep brief; includes "likely objections" and rapport builders.
  • competitor-intel: battlecard format with landmine questions, useful when role-playing "why not [competitor]" moments.

Verdict

Steal the 30-minute call clock and the token-isolation / subagent discipline as engineering patterns; the skills themselves overlap with zubair's.

🧩

Synthesis: adapt vs build for the 8-skill architecture

The decision table
Planned skillVerdictSource material
/roleplayBuildThe engine, personas, in-character rules, and audible loop are all custom. Adapt content: louisblythe discovery + closing + pricing-negotiation for scenario scripts; sales-psychology + building-rapport personality types for persona realism; northrn's 30-minute call clock for session structure; marketingskills persona cards as the persona template.
/objection-gymAdapt-heavyClosest to assemble-from-parts. zubair's sales-objections 15-objection corpus plus FFF and ABC frameworks; louisblythe LAER plus 3-Response Drill and Pause Practice as drill mechanics; marketingskills two-format output (drill, then a personal quick-reference card).
/daily-repAdaptHarvest the Practice Exercises sections across the louisblythe Core 20 (objection journal, silence game, direct-ask practice, summary challenge, question-ratio tracking) into a rotating drill deck.
/fear-ladderBuildNothing like it exists anywhere. Only louisblythe resilience contributes (reframes, recovery routine, rejection-as-data). Graduated exposure design is from scratch.
/call-reviewAdaptlouisblythe conversation-quality-scoring dimensions and weights; deal-review-win-loss debrief questions; active-listening question-ratio metric.
/pitch-polishAdaptmarketingskills sales-enablement talk-track structure; louisblythe dynamic-script-generation component approach; she already has offers and positioning skills feeding this.
/certifyBuildNo certification or gating pattern exists in any repo. Borrow the evidence-based 0-100 scoring shape from zubair sales-qualify; pass thresholds and stage-gate design are custom.
/scorecardAdaptlouisblythe conversation-quality-scoring weighted rubric plus human-in-the-loop-training improvement tracking; zubair's scoring-with-evidence output format.

Top recommendations

  1. Install the louisblythe Core 20 as knowledge only (security-review pass first, per standing rule). They're reference curriculum, not runnable trainers. Never bulk-install the other 102.
  2. Build /roleplay first and from scratch. It's your stated top want and the one thing no repo provides. Seed each session with a scenario brief generated from louisblythe frameworks plus your real business context.
  3. Mine zubair's sales-objections file immediately into an objections/ corpus for /objection-gym: 15 universal plus Vitalency / TrainerBooked / SEO-specific ones generated with its industry rules.
  4. Adopt the weighted scoring rubric (discovery-info weights: pain 25, budget 20, timeline 20, process 15, stakeholders 15, current solution 10) as the v1 /call-review and /scorecard backbone. Tune weights to your context rather than inventing from zero.
  5. Copy northrn's engineering discipline (tool restriction, token isolation, model-routing notes, dynamic tuning sections) as the SKILL.md house style for everything we build.

Coverage: louisblythe 122 inventoried / 14 deep-read · marketingskills 47 inventoried / 1 deep-read · zubair 14 inventoried / 5 deep-read · sales-skills/sales structurally sampled (confirmed skip) · northrn 7 inventoried / 3 deep-read.

Compiled July 12, 2026 · ← Hub