| name | sales-prep |
| description | Meeting Preparation Brief Engine — generates MEETING-PREP.md with 11 fixed sections covering company research, attendee intelligence, competitive context, and tactical preparation. Invoked standalone via `/sales prep <url>`. Industry-agnostic engine: section structure, question ordering, and response frameworks are constants; all prospect-specific content is runtime variable.
|
Meeting Preparation Brief
Tier 0 Doctrine
- Tier: Master (standalone skill, invoked directly)
- Authority: Operates on user-supplied URL + optional attendee/meeting context; produces one output artifact
- Determinism first: All section structure, question ordering, framework templates, and agenda layouts are fixed constants. LLM assembles research into deterministic slots — never interprets structure.
- No fabrication: Every factual claim must cite a source URL or be labeled inference. Absence of data is disclosed, not invented.
IMO (Top-Level)
| Layer | Responsibility |
|---|
| Ingress | Prospect URL + optional attendee names, meeting date/time, meeting purpose, product context (schema validation only) |
| Middle | 4-phase research (company, attendee, competitive, industry); assembly into 11 fixed sections + agenda templates |
| Egress | MEETING-PREP.md written to working directory (read-only structured output) |
Constants
| Constant | Value | Authority |
|---|
| Output File | MEETING-PREP.md | Fixed |
| Output Sections | 11 fixed sections (see Workflow) | Fixed |
| Cheat Sheet Items | 5 items + opening line + key question + trap to avoid | Fixed |
| Talking Points Count | 5-7 per brief | Fixed |
| Discovery Questions Count | 10 per brief | Fixed |
| Discovery Question Order | rapport (2), situation (2), pain (2), impact (2), future (2) | Fixed |
| Objections Count | 5 per brief | Fixed |
| Objection Framework | Feel-Felt-Found | Fixed |
| Success Metric Tiers | minimum, target, stretch | Fixed |
| Next Step Tiers | bold, standard, minimum | Fixed |
| Agenda Templates | 30-minute and 60-minute | Fixed |
| Research Phases | Company, Attendee, Competitive, Industry | Fixed; 4 phases |
| Company Research Surfaces | Homepage, About, Products, Blog/News, Careers, Case Studies | Fixed |
| Attendee Profile Fields | Background, Recent Activity, Communication Style, Likely Priorities, Rapport Anchors, How to Win | Fixed |
| Communication Styles | detail-oriented, big-picture, data-driven, relationship-focused | Fixed |
| Attendee Roles (decision map) | decision-maker, influencer, gatekeeper | Fixed |
| Competitive Intel Categories | Current solutions, switching triggers, what NOT to say, differentiation | Fixed |
| Talking Point Format | Statement/Question + Context + Leads-to | Fixed |
| Discovery Question Format | Question + Purpose + Expected Response + Follow-Up + Listen For | Fixed |
| Objection Format | Exact words + Hidden concern + Feel-Felt-Found response + Proof point | Fixed |
| Landmine Format | Topic + Why avoid + If-it-comes-up handling | Fixed |
| Next Step Format | Exact wording + date suggestion + pushback response | Fixed |
Variables
| Variable | Source | Runtime |
|---|
target_url | User invocation (/sales prep <url>) | Per-run |
attendee_names | User input (optional) | Per-run |
meeting_date | User input (optional) | Per-run |
meeting_purpose | User input (optional) | Per-run |
product_context | User input (optional) | Per-run |
company_name | Homepage extraction | Discovered |
company_description | Homepage, About page | Discovered |
company_size | About page, LinkedIn | Discovered |
funding_stage | WebSearch results | Discovered |
revenue_estimate | WebSearch results | Discovered |
growth_signals | Careers page, news | Discovered |
product_offerings | Products/Services pages | Discovered |
recent_news | WebSearch results | Discovered |
open_roles | Careers page | Discovered |
case_studies | Testimonials page | Discovered |
attendee_profiles | LinkedIn, public presence | Discovered |
decision_dynamics | Role analysis | Computed |
current_solutions | Website tech, job posts, integrations | Discovered |
competitor_advantages | Competitive analysis | Computed |
industry_trends | WebSearch results |
Workflow
BLOCK 1: Research Collection
Governed by: C&V
Constants: 4 research phases, 6 company research surfaces, attendee profile fields, competitive intel categories, prior analysis file list
Variables: target_url, attendee_names, company_name, company_description, company_size, funding_stage, revenue_estimate, growth_signals, product_offerings, recent_news, open_roles, case_studies, attendee_profiles, current_solutions, industry_trends, industry_pain_points, prior_analysis
IMO:
- Ingress:
target_url validated as URL; optional inputs validated if present; check working directory for prior analysis files
- Middle:
- Phase 1 — Company: WebFetch homepage, about, products, blog/news, careers, case studies. WebSearch
"[company]" news [year] and "[company]" funding OR acquisition. Extract: description, value prop, team size, locations, pricing model, content themes, hiring signals, customer logos.
- Phase 2 — Attendee: If names provided: search LinkedIn profiles, recent posts, conference talks, published articles. Build per-person profile using 6 fixed fields. Predict communication style from role/seniority. If no names: infer likely attendees from leadership page + meeting purpose + deal stage; label as "Predicted Attendees" with confidence level.
- Phase 3 — Competitive: Detect current tools from website scripts, meta tags, job post requirements, partner pages, social mentions. For each competitor: note usage, identify advantages, flag topics to avoid.
- Phase 4 — Industry: WebSearch for 2-3 recent trends/challenges. Identify regulatory changes, market shifts, competitive pressures.
- If prior analysis files exist, ingest findings — do not re-research covered ground.
- Run fetches in parallel where possible.
- Egress: Structured research data for all 4 phases; source URL recorded for every factual claim
Go/No-Go: Proceed if homepage fetch succeeds. If homepage unreachable, halt with error. Partial data from other phases is acceptable — flag gaps, continue.
BLOCK 2: Brief Assembly
Governed by: IMO
Constants: 11 section structure, cheat sheet format (5 items + 3 elements), talking point format, discovery question format and ordering (rapport/situation/pain/impact/future x2), objection format (Feel-Felt-Found), success metric tiers (min/target/stretch), landmine format, next step tiers (bold/standard/minimum) with exact-wording requirement
Variables: All research variables from Block 1 + cheat_sheet, talking_points, discovery_questions, objection_responses, success_metrics, competitive_landmines, next_steps
IMO:
- Ingress: All structured research data from Block 1
- Middle:
- Section 1 — Cheat Sheet: Synthesize 5 most important facts. Generate personalized opening line, key question, trap to avoid. This section must stand alone — a reader of only the cheat sheet walks in meaningfully prepared.
- Section 2 — Company Snapshot: One paragraph (4-6 sentences) + quick-reference table (10 fields: Company, Website, Industry, Founded, Employees, Revenue, Funding, HQ, Key Products, Target Market).
- Section 3 — Attendee Profiles: Per-attendee block with 6 fixed fields. If multiple attendees: add decision dynamics (decision-maker/influencer/gatekeeper) and meeting politics.
- Section 4 — Business Situation: Current state, recent changes, growth trajectory, 3-5 key challenges, addressable opportunities. Frame challenges as opportunities — "room for improvement" not "problems."
- Section 5 — Competitive Context: Current solutions with satisfaction assessment, switching triggers, topics to avoid, 2-3 differentiation points specific to this prospect.
- Section 6 — Talking Points (5-7): Each must reference specific research. Format: statement/question + context + leads-to. Never generic ("How's business?", "Tell me about your company", "What keeps you up at night?" are banned).
- Section 7 — Discovery Questions (10): Ordered: Q1-2 rapport, Q3-4 situation, Q5-6 pain, Q7-8 impact, Q9-10 future/decision. Each has: question, purpose, expected response, follow-up, listen-for signals. Questions must be genuinely curious, not leading.
- Section 8 — Objections (5): Most likely objections for this prospect's size, industry, solutions, stage. Each: exact words, hidden concern, Feel-Felt-Found response with metrics, proof point.
- Section 9 — Success Metrics: Minimum (bare-minimum worthwhile outcome), target (realistic best), stretch (ideal). Each specifies: what prospect says/agrees to, concrete next step established, information obtained.
- Section 10 — Competitive Landmines: Topics to avoid. Each: topic, why avoid, graceful handling if raised.
- Section 11 — Next Steps: Bold ask (most ambitious), standard ask (reasonable), minimum ask (fallback). Each: exact wording, specific date/time suggestion, pushback response.
- Egress: Complete 11-section brief content assembled in memory
Go/No-Go: All 11 sections must be populated. Any section with insufficient data must state what is missing and why. No empty sections permitted.
BLOCK 3: Agenda Generation
Governed by: Circle
Constants: 30-minute agenda blocks (5 segments), 60-minute agenda blocks (6 segments), time allocations per segment
Variables: meeting_purpose, agenda, talking_points, discovery_questions
IMO:
- Ingress: Brief content from Block 2 + meeting context
- Middle:
- Generate 30-minute agenda: Rapport (0:00-2:00) with specific talking point, Discovery (2:00-10:00) with top 4 questions, Value/Demo (10:00-20:00) with key points, Q&A (20:00-25:00) with likely questions, Next Steps (25:00-30:00) with proposed step and date.
- Generate 60-minute agenda: Rapport (0:00-5:00), Discovery (5:00-20:00) with all 10 questions prioritized, Presentation (20:00-40:00) tailored to stated needs, Discussion (40:00-50:00) with proof points, Pricing (50:00-55:00) if appropriate for stage, Next Steps (55:00-60:00).
- Cross-reference agenda activities to specific brief sections.
- Egress: Two agenda tables with prospect-specific notes per segment
Go/No-Go: Both agenda templates must be produced regardless of whether meeting duration is known.
BLOCK 4: Output & Source Audit
Governed by: CTB
Constants: Output file name (MEETING-PREP.md), document header format, section numbering, research sources section
Variables: All assembled content from Blocks 2-3, all source URLs collected in Block 1
IMO:
- Ingress: Complete brief content + agenda tables + source URL registry
- Middle:
- Assemble final document with header: title, generated date, meeting date (if provided), meeting purpose (if provided), prepared-by line.
- Write 11 sections with horizontal rule separators, numbered per constant structure.
- Append Suggested Agenda section with both templates.
- Append Research Sources section listing every URL fetched and source consulted.
- Verify every factual claim has a source citation or is labeled as inference.
- Write
MEETING-PREP.md to working directory.
- Egress:
MEETING-PREP.md written to disk; confirmation message to user
Go/No-Go: File must be written successfully. If write fails, report error. Output always produced — partial data sections are disclosed, never omitted.
Rules
- Never include generic content. Every talking point, question, objection, and recommendation must reference specific research about this prospect.
- Never fabricate evidence. Every factual claim cites a source URL. Speculation is labeled as inference, never presented as fact.
- Never use manipulative tactics. No NLP tricks, psychological pressure techniques, or leading questions designed to force conclusions.
- Never bash competitors. Competitive references are factual. Unverifiable claims are labeled "commonly reported" or omitted entirely.
- Never produce empty sections. If data is insufficient, state what is missing and the impact on preparation quality.
- Never re-research covered ground. If prior analysis files exist in the working directory, incorporate their findings directly.
- Never exceed the fixed section count. Output has exactly 11 sections + agenda + sources. No additions, no removals, no reordering.
- Never reorder discovery questions. The sequence rapport-situation-pain-impact-future is a constant, not a suggestion.
- Never omit the Cheat Sheet opening line, key question, and trap. These three elements are mandatory even when research is sparse.
- Never propose next steps without exact wording. Each tier (bold/standard/minimum) includes the literal sentence the salesperson should say.
Reference Pointers
| Reference | Location |
|---|
| Prospect analysis skill | skills/sales-prospect/SKILL.md |
| Qualification skill (BANT + MEDDIC) | skills/sales-qualify/SKILL.md |
| Opportunity assessment subagent | agents/sales-opportunity.md |
| ICP definition | IDEAL-CUSTOMER-PROFILE.md (working directory, optional) |
| Prior analysis files | PROSPECT-ANALYSIS.md, COMPANY-RESEARCH.md, LEAD-QUALIFICATION.md, COMPETITIVE-INTEL.md, DECISION-MAKERS.md (working directory, optional) |
| Doctrine | templates/doctrine/ARCHITECTURE.md (IMO, Hub-Spoke, CTB) |
| Skill creation rules | skills/skill-creator/SKILL.md |
