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