| name | sales-competitors |
| description | Competitive Intelligence Engine Subagent — detects what tools/solutions a prospect currently uses via 6 detection methods, builds 10-section battle cards per overlapping competitor, and assembles a scored competitive strategy. Trigger standalone via `/sales competitors <url>` producing COMPETITIVE-INTEL.md, or as subagent during `/sales prospect <url>` returning Competitive Position Score 0-100. Industry-agnostic engine.
|
| weight | 0.15 |
| tier | spoke |
Competitive Intelligence Engine
Tier 0 Doctrine
- Tier: Spoke (subagent of sales-prospect orchestrator; also standalone via
/sales competitors <url>)
- Authority: Reads prospect URL from orchestrator or user; detects current solutions and builds competitive battle cards only
- Determinism first: All detection uses 6 declared methods. All scoring uses declared formulas and confidence levels. No LLM interpretation of scores.
- No fabrication: Every competitor capability must have a source. Unverified claims are labeled "Reported." Absence of data is noted, not invented.
IMO (Top-Level)
| Layer | Responsibility |
|---|
| Ingress | Company URL + optional prior analysis files (PROSPECT-ANALYSIS.md, COMPANY-RESEARCH.md, LEAD-QUALIFICATION.md) — schema validation only |
| Middle | 6-method solution detection; 10-section battle card generation per competitor; feature gap analysis; win/loss pattern recognition; competitive strategy assembly; 0-100 scoring |
| Egress | COMPETITIVE-INTEL.md (standalone) or Competitive Position Score 0-100 with structured data (subagent) — read-only output |
Constants
| Constant | Value | Authority |
|---|
| Detection Methods | 6 fixed: Website Technology Analysis, Job Posting Analysis, Case Study/Partnership Search, Review Site Search, Tech Stack Detection Services, Social Signal Analysis | Locked |
| Confidence Levels | High (explicit mention/badge), Medium (script tag/indirect reference), Low (inference from job posting/industry norm) | Fixed; 3 levels |
| Battle Card Sections | 10 fixed: What They Use It For, Strengths (3-5), Weaknesses (3-5), Your Advantages + proof, Their Advantages + Neutralization, Switching Cost Assessment (5 factors), Switching Triggers (5), Landmine Questions (5), Trap to Avoid, Positioning Statement | Locked |
| Switching Cost Factors | 5 fixed: Technical migration, Financial cost, Organizational change, Data portability, Timeline estimate | Fixed |
| Feature Comparison Rules | Include wins AND losses; specific details not checkmarks; mark unverified as "Reported"; organize by prospect importance | Fixed |
| Win/Loss Pattern Format | 3-5 win reasons + 3-5 loss reasons per competitor | Fixed |
| Deal Qualification Signal Format | "If prospect says X -> favors you/competitor" | Fixed |
| Competitive Strategy Structure | Strategy Summary, Conversation Sequence (5 topics), What to Lead With, What to Avoid, Displacement Timeline | Fixed; 5 sections |
| Scoring Sub-Dimensions | Detection Breadth, Battle Card Depth, Strategy Coherence, Evidence Quality, Displacement Feasibility | Fixed; 5 sub-dimensions |
| Sub-Dimension Range | 0-20 each | Fixed |
| Score Formula | sum of 5 sub-dimensions | Produces 0-100 |
| Output: Standalone | COMPETITIVE-INTEL.md | Fixed |
| Output: Subagent | Competitive Position Score 0-100 + structured data block | Fixed |
| Output Sections | Executive Summary, Current Solutions Detected, Battle Cards, Feature Gap Analysis, Win/Loss Patterns, Competitive Positioning Statements, Switching Cost Assessment, Recommended Strategy, Detection Sources | Fixed; 9 sections |
| Refresh Recommendation | Every 3-6 months or before major competitive deal | Fixed |
Variables
| Variable | Source | Runtime |
|---|
target_url | User input or orchestrator briefing | Per-run |
company_name | Homepage detection | Discovered |
invocation_mode | standalone or subagent | Per-run |
prior_analysis | PROSPECT-ANALYSIS.md, COMPANY-RESEARCH.md, LEAD-QUALIFICATION.md (if present) | Per-run |
detected_solutions | 6 detection methods | Discovered |
confidence_ratings | Per solution — High/Medium/Low | Computed |
solution_categories | CRM, Marketing Automation, Analytics, Customer Support, Project Management, Communication, etc. | Discovered |
overlapping_competitors | Subset of detected solutions that compete with offering | Computed |
battle_cards | 10-section card per overlapping competitor | Computed |
feature_comparison | Side-by-side table across competitors | Computed |
win_loss_patterns | 3-5 win reasons + 3-5 loss reasons per competitor | Computed |
deal_qualification_signals | Computed from win/loss patterns | Computed |
competitive_strategy | Strategy Summary + Conversation Sequence + Lead/Avoid + Displacement Timeline | Computed |
detection_breadth_score | Sub-dimension 0-20 | Computed |
battle_card_depth_score | Sub-dimension 0-20 | Computed |
strategy_coherence_score | Sub-dimension 0-20 | Computed |
evidence_quality_score | Sub-dimension 0-20 | Computed |
|
Workflow
BLOCK 1: Solution Detection
Governed by: C&V
Constants: 6 detection methods, 3 confidence levels, source priority
Variables: target_url, company_name, detected_solutions, confidence_ratings, prior_analysis
IMO:
- Ingress: URL validated as reachable; prior analysis files loaded if present in working directory
- Middle:
- Execute all 6 detection methods in parallel via WebFetch and WebSearch:
- Website Technology Analysis: Fetch prospect URL. Inspect HTML source for script tags, meta tags, link tags referencing known tools. Check homepage, footer, integrations page, partners page for tool logos/badges. Detect "Powered by" footers. Identify website platform, hosting, CDN, email provider. Check login/portal links and API documentation pages.
- Job Posting Analysis: WebSearch
"[Company Name]" careers, "[Company Name]" open jobs, "[Company Name]" hiring. Extract tool-specific requirements, tech stack mentions, vendor certifications, process tool references. These are HIGH confidence signals.
- Case Study/Partnership Search: WebSearch
"[Company Name]" case study [competitor], "[Company Name]" customer story, "[Company Name]" testimonial. Check if prospect appears on competitor websites as customer, in testimonials, as partner.
- Review Site Search: WebSearch
site:g2.com "[Company Name]", site:capterra.com "[Company Name]", "[Company Name]" review [tool category]. Note which products reviewed, satisfaction level, specific complaints/praise, review date.
- Tech Stack Detection Services: WebSearch
site:builtwith.com [Company Domain], site:stackshare.io "[Company Name]", "[Company Domain]" technology stack. Catalog technologies not visible through direct inspection.
- Social Signal Analysis: WebSearch
"[Company Name]" [tool category] site:linkedin.com, "[Company Name]" "we use" OR "switched to" OR "implemented". Look for blog posts, LinkedIn posts, conference talks, community forum posts about tool choices.
- For each detected technology, record: tool/platform name, evidence source (URL/page element/signal), confidence level (High/Medium/Low)
- If prior analysis files exist, incorporate prospect priorities and pain points into detection weighting
- If URL unreachable: attempt www/non-www and https/http variants; if still unreachable, halt
- Egress: Structured detection inventory: tool name + evidence + confidence per entry
Go/No-Go: Proceed if at least 1 solution detected. If zero solutions detected across all 6 methods, halt and report to user.
BLOCK 2: Battle Card Generation
Governed by: IMO
Constants: 10 battle card sections, switching cost factors (5), feature comparison rules
Variables: overlapping_competitors, battle_cards, feature_comparison, solution_categories
IMO:
- Ingress: Full detection inventory from Block 1
- Middle:
- Categorize all detected solutions into relevant categories (CRM, Marketing Automation, Analytics, Customer Support, Project Management, Communication, etc.)
- Identify which detected solutions overlap with or compete against offering — these receive full battle cards
- For EACH overlapping competitor, build 10-section battle card:
- What They Use It For — specific description based on detection evidence
- Strengths (3-5) — genuine strengths with why each matters to prospect. Credibility requires honesty.
- Weaknesses (3-5) — specific weaknesses with evidence. "Bad support" is too generic; "48-hour average ticket response, no dedicated AM under $50K ARR" is specific.
- Your Advantages — each with proof point (metric, case study, testimonial)
- Their Advantages + Neutralization — honest about where competitor wins; for each, explain how to neutralize or reframe
- Switching Cost Assessment — 5 factors table: Technical migration (Low/Med/High + details), Financial cost (Low/Med/High + details), Organizational change (Low/Med/High + details), Data portability (Easy/Moderate/Difficult + details), Timeline estimate (weeks/months + details)
- Switching Triggers (5) — events that commonly cause customers to switch (contract renewal price increase, outgrowing tier structure, feature gap blocking workflow, new leadership vendor consolidation, poor support during critical incident)
- Landmine Questions (5) — questions exposing competitor weaknesses without bashing. Must be genuinely curious questions any smart buyer would ask.
- Trap to Avoid — what NOT to say about this competitor and why
- Positioning Statement — one sentence: "While [Competitor] is [honest strength], [Your Company] [differentiator] which means [specific outcome for prospect]."
- Build feature comparison table: side-by-side across competitors. Include features where you win AND lose. Use specific details not checkmarks. Mark unverified as "Reported." Organize by prospect importance.
- Egress: Structured battle cards + feature comparison table per competitor
Go/No-Go: Proceed if at least 1 battle card produced. If no overlapping competitors found, produce detection report only (no battle cards needed).
BLOCK 3: Win/Loss Analysis & Strategy
Governed by: Circle
Constants: Win/loss pattern format (3-5 each), deal qualification signal format, competitive strategy structure (5 sections)
Variables: win_loss_patterns, deal_qualification_signals, competitive_strategy
IMO:
- Ingress: All battle cards + feature comparison from Block 2 + prior analysis context
- Middle:
- For each competitor, compile win/loss patterns:
- Win Patterns (3-5): "We win when [specific situation or buyer profile]" — derived from battle card analysis, switching triggers, and feature advantages
- Loss Patterns (3-5): "We lose when [specific situation or buyer profile]" — derived from competitor strengths and their advantages
- Generate deal qualification signals from win/loss patterns: "If prospect says [X] -> favors you because [reason]", "If prospect prioritizes [Z] -> strong deal for you", "If prospect prioritizes [W] -> consider deprioritizing"
- Assemble competitive strategy:
- Strategy Summary: One paragraph — which competitor to focus on displacing, which to ignore, what messaging to lead with
- Conversation Sequence: 5 topics in recommended order, each with reasoning
- What to Lead With: Single strongest differentiator with talking point script
- What to Avoid: Topics, claims, comparisons to deliberately avoid with reasons
- Displacement Timeline: When to start (relative to contract renewal), expected evaluation period, migration timeline, full transition period
- Validate: every claim traceable to detection evidence or battle card analysis. No unsourced assertions.
- Egress: Validated win/loss patterns + deal qualification signals + competitive strategy
Go/No-Go: Proceed unconditionally. If limited data, reduce strategy confidence and note gaps.
BLOCK 4: Scoring & Output Assembly
Governed by: CTB
Constants: 5 scoring sub-dimensions, sub-dimension range 0-20, score formula, output sections (9), refresh recommendation
Variables: All 5 sub-dimension scores, competitive_position_score, invocation_mode
IMO:
- Ingress: All data from Blocks 1-3: detection inventory, battle cards, feature comparison, win/loss patterns, competitive strategy
- Middle:
- Score 5 sub-dimensions (0-20 each):
- Detection Breadth (0-20): How many methods yielded results; how many solutions detected; confidence distribution. 6/6 methods with High confidence = 18-20; 1-2 methods with Low confidence = 2-6.
- Battle Card Depth (0-20): Completeness of 10 sections per card; specificity of evidence; actionability of talking points. All 10 sections with specific evidence = 16-20; sparse or generic = 4-8.
- Strategy Coherence (0-20): Logical consistency between detection, battle cards, and strategy; clear prioritization; actionable conversation sequence. Fully coherent = 16-20; contradictory or vague = 4-8.
- Evidence Quality (0-20): Proportion of High confidence detections; source diversity; recency of sources. Mostly High + multiple sources = 16-20; mostly Low + single source = 2-6.
- Displacement Feasibility (0-20): Realistic switching costs; identified triggers aligned with prospect context; viable timeline. Clear path with triggers = 16-20; high switching cost + no triggers = 2-6.
- Compute
competitive_position_score = sum of 5 sub-dimensions
- If
invocation_mode = standalone: assemble full COMPETITIVE-INTEL.md with all 9 output sections (Executive Summary, Current Solutions Detected, Battle Cards, Feature Gap Analysis, Win/Loss Patterns, Competitive Positioning Statements, Switching Cost Assessment, Recommended Strategy, Detection Sources)
- If
invocation_mode = subagent: return structured data block with Competitive Position Score, sub-dimension breakdown, top competitors, key strategy points
- Append refresh recommendation: "Refresh every 3-6 months or before major competitive deal"
- Note detection source dates throughout
- Egress: COMPETITIVE-INTEL.md written to disk (standalone) or structured score block returned (subagent)
Go/No-Go: Output is always produced. Clearly note all data gaps and confidence limitations.
Rules
- Never fabricate competitor capabilities. If a feature, price, or capability cannot be verified, label it as "Reported" or "Unverified." Sales credibility depends on accuracy.
- Never omit competitor strengths from a battle card. Every battle card must include genuine strengths. One-sided portrayals destroy credibility the moment the prospect pushes back.
- Never recommend bashing competitors. Battle cards position and differentiate. Negative selling backfires.
- Never use generic assessments. "Better support" is useless. "Dedicated AM for all plans, 2-hour avg response vs their 24-hour SLA" is actionable. Specific over generic always.
- Never skip confidence labeling on detected technologies. High/Medium/Low must accompany every detection. A badge on the website carries different weight than an industry-norm inference.
- Never underestimate or overestimate switching costs. Underestimating looks naive. Overestimating makes the deal feel impossible. Accuracy matters.
- Never ignore prior analysis files when present. If PROSPECT-ANALYSIS.md, COMPANY-RESEARCH.md, or LEAD-QUALIFICATION.md exist, incorporate their findings into competitive positioning.
- Never write transparent trap questions. Landmine questions must be genuinely curious questions any smart buyer would ask, not obvious traps designed to embarrass the competitor.
- Never omit refresh recommendation date. Competitive intelligence has a shelf life. Note source dates and recommend refresh timeline.
- Never present one-sided feature comparisons. Include features where you win AND where you lose. Credibility requires showing both.
Reference Pointers
| Reference | Location |
|---|
| Orchestrator | skills/sales-prospect/SKILL.md |
| Company research skill | skills/sales-research/SKILL.md |
| Qualification skill | skills/sales-qualify/SKILL.md |
| Contacts skill | skills/sales-contacts/SKILL.md |
| Prior analysis files | PROSPECT-ANALYSIS.md, COMPANY-RESEARCH.md, LEAD-QUALIFICATION.md (working directory, optional) |
| Doctrine | templates/doctrine/ARCHITECTURE.md (IMO, Hub-Spoke, CTB) |
| Skill creation rules | skills/skill-creator/SKILL.md |
