v1.10.0

Changes

• Added optional guidance for reviewed social source packets in startup-competitors.
• Limited social packets to customer language, community/social signals, source labels, URLs, and timestamps unless primary sources confirm stronger factual claims.
• Clarified that funding, hiring, and pricing claims still require primary-source confirmation.

Assets

• startup-competitors.zip

v1.9.0 - Validation-First Flow & Pitch Deck Outline

What's new

startup-design: validation-first flow (#2, thanks @llillo)

• Pre-Flight Check (Phase 0.5) - a 5-minute sanity scan before intake: dominant incumbents, prior failures of similar ideas, regulatory blockers. Context, not a verdict.
• Customer Discovery gate (Phase 3.7) - real customer interviews between the Research Gate and Strategy, with a full Mom Test-based protocol in the new references/customer-interview.md (who to interview, 25-30 min question structure, notes template, synthesis guide, go/no-go criteria). The gate is explicitly deferrable: the founder can continue on research data alone, with consequences stated plainly.
• Two-stage financial model - every financial file declares its stage: Stage A (assumption-based, default) or Stage B (evidence-based, anchored to interview signals). Plus an assumption sensitivity table in projections.
• Final Assessment Dashboard now shows Customer Interviews (count + confirmation rate) and Financial Stage.

startup-pitch: deck outline (idea from #3, thanks @Haoyang0708)

• New deliverable pitch-deck-outline.md - slide-by-slide deck structure (10-12 slides) derived from the full narrative: takeaway headlines (never labels), body bullets under 30 words, visual suggestions, speaker notes per slide, and a deck quality checklist.

All skills

• Resume Check (Phase 0) added to startup-competitors, startup-positioning, and startup-pitch - interrupted sessions now resume from PROGRESS.md in every skill, not just startup-design.
• Flow diagrams now correctly describe waves as sequential (agents within a wave run in parallel).
• Reference tables fixed and updated with accurate line counts; tables of contents added to the large synthesis references.
• startup-design SKILL.md restructured under the 500-line limit: intake questions moved to references/intake-questions.md, Strategy/Brand/Product/Validation file specs to references/output-specs.md.

Install / Update

Claude Code plugin:

claude plugin update startup@startup-skill

Claude.ai: download the .skill files below and upload in Settings -> Skills.

v1.8.0 - Deep-reasoning cues for new models

What's new

Prompting-style optimization for the latest Claude models (Opus 4.7 / Sonnet 4.6 / Haiku 4.5).

Added concise, reasoned guidance at each skill's analytical crux, where the model should reason deliberately before writing. On models that support extended thinking, these cues engage it at the highest-leverage moments:

• startup-design - synthesis (raw research to deliverables) + go/no-go research gate
• startup-competitors - synthesis / battle cards
• startup-positioning - Onliness Test / Dunford 5+1 derivation
• startup-pitch - pitch construction / element ordering

Surgical additions, no rewrites. The skills keep their existing structure and behavior.

Install

Download the .skill file for the skill you want and install it, or add the plugin marketplace from the repo.

v1.7.0 — Verification Agent & Dynamic Research Scaling

What's New

Verification Agent (all 4 skills)

Post-synthesis V1: Verification agent that audits all deliverables for:

• Unlabeled claims (missing [Data]/[Estimate]/[Assumption]/[Opinion] tags)
• Internal contradictions across files
• Confidence rating consistency (does the evidence match the rating?)
• Missing data gaps and flags sections
• Stale data and duplicate-source false corroboration
• Skill-specific checks: cross-phase consistency (design), battle card coherence (competitors), positioning statement vs. research (positioning), cross-format pitch consistency (pitch)

Critical issues pause the flow for user review. Produces verification-report.md.

Dynamic Research Scaling (all 4 skills)

Complexity scoring (3-9) after intake determines research depth:

• Light (3-4): merged agents, 2-3 search rounds — quick scan for niche markets
• Standard (5-7): current behavior, unchanged
• Deep (8-9): extra agents per wave, 5-6 search rounds — thorough due diligence

Users see the complexity assessment and can override with light, deep, or ok.

Files Changed

• 8 new reference files (verification-agent.md + research-scaling.md per skill)
• 4 SKILL.md files updated with new phases
• 4 research-synthesis.md files updated with verification step
• AGENTS.md, CLAUDE.md, GUIDELINES.md updated
• marketplace.json bumped to 1.7.0

v1.6.0

What's New

Final Assessment Dashboard — After all 8 phases complete, startup-design now prints a visual summary dashboard directly in the conversation with all key findings:

• Idea, Founder Fit, Market (TAM + growth + maturity), Competitors (count + funding landscape), Positioning, Business Model, MVP Scope, Model Viability, Top Risk, Score, and Verdict

This gives founders an instant visual payoff before diving into the 30+ generated files.

v1.5.0 — Startup Pitch

What's New

startup-pitch (NEW)

Build investor-ready pitch content in 6 formats using a structured 8-element framework:

• pitch-full.md — Full 10-minute narrative with speaker notes
• pitch-5min.md — Compressed 5-minute version
• pitch-2min.md — Word-for-word verbal pitch script
• pitch-1min.md — Elevator pitch (formal + casual)
• pitch-email.md — Investor cold email + follow-up template
• pitch-appendix.md — Top 10 Q&A, objection handling, known weaknesses
• pitch-scorecard.md — Scoring rubric (0-80) with investor-readiness verdict

Features:

• 2 research waves (audience intelligence + competitive framing)
• 4 pitch ordering templates (traction-led, team-led, insight-led, pre-traction)
• Investor roleplay practice with 4 personas (angel, seed VC, Series A VC, accelerator)
• Integrates with prior startup-design/competitors/positioning output
• Honesty protocol: flags inflated metrics, weak team slides, vague asks

Other Changes

• Updated README, CLAUDE.md, CONTRIBUTING.md with startup-pitch documentation
• Updated marketplace.json with new skill entry

Full Changelog: v1.4.1...v1.5.0

v1.4.1 — README Banner & Badges

What's Changed

• Added hero banner image to README
• Added version, license, and stars badges
• Collapsible installation and updating sections
• Claude Code Plugin install commands always visible

Full Changelog: v1.4.0...v1.4.1

v1.4.0 — Messaging Implications + README Refresh

What's Changed

• New deliverable: messaging-implications.md in startup-positioning — messaging hierarchy, category vs customer language map, words to use/avoid, social proof guidelines, freemium positioning
• Enhanced: positioning-statement.md template now includes tagline stress-testing, one-liner variants, and freemium positioning
• Improved: README copy — clearer, more direct descriptions of each skill
• Updated: research-synthesis.md with category/customer language check and value anchor separation

Full Changelog: v1.3.0...v1.4.0

v1.3.0 — Startup Positioning + Cross-Skill Audit

What's New

New Skill: startup-positioning

Market positioning strategy using April Dunford's framework, enriched with JTBD discovery, Moore positioning statement, and Neumeier's Onliness Test. Produces:

• Positioning document (Dunford's 5+1 components)
• Positioning statements (Moore + Neumeier + elevator pitch)
• Competitive alternatives map (JTBD-informed)
• Market category analysis

Cross-Skill Improvements

Complete audit across all 3 skills — 25 fixes applied:

• Honesty protocol now in all skills (competitors was missing it)
• Agent naming standardized to A1/B1/C1 convention everywhere
• 4 data labels ([Data], [Estimate], [Assumption], [Opinion]) in all skills
• Post-research checkpoint before synthesis in all skills
• PROGRESS.md tracking in all skills
• Standardized headers (Skill: startup-{name} | Generated: {date}) in all output templates
• Confidence ratings explicit in all deliverable templates
• Fixed broken reference in competitors synthesis phase
• Fixed agent count and framework timing in design
• Added switching cost analysis and adjacent solutions to output templates

Benchmark Results (old vs new)

Skill	Before	After
startup-competitors	37.5%	87.5%
startup-design	37.5%	62.5%

.skill files

All 3 skills included as .skill packages for Claude.ai installation.

Full Changelog: v1.2.1...v1.3.0

v1.2.1 — Improved docs and .skill packages

What's Changed

• Rewritten README with benefit-focused copy
• .skill files included as release assets for easy Claude.ai installation

Full Changelog: v1.2.0...v1.2.1