Zhang Lu.
← Skills

/boss

Zero-dependency onboarding: install it and run `/boss <topic>` for a verdict deliberated independently by 5 judges — Phase 0 routing → 5-dimension parallel research → Lead synthesis → 1 anchor (baseline conviction) + 4 dimension judges scoring independently → panel_summary with anchor_delta + 30/90/365 attribution checkpoints. Unlike the single-lens `*-perspective` skills, `/boss` is a **full pipeline orchestration**: - 1 anchor judge (anchor mental_models: trigger-threshold-action / step-skip detection / counter-party elimination) - 4 dim judges auto-selected per topic (panel auto-select: strategic/customer/organizational/product/brand/financial/cross_domain) - 5-lens scoring (reasoning / evidence / counter / falsifiability / real_world_resilience) - adversarial_view three fields enforced (CLAUDE.md §4.5) - Versioned freeze (chmod 444) + Wiki back-references Router behavior is built in: - If `reports/<brand-slug>/report.md` already exists → EVOLUTION mode (diff plan + partial re-run) - Otherwise → FRESH mode (full Phase 1-5) Trigger patterns: - `/boss <topic or brand-slug>` — standard invocation - `/boss <topic> --quick` — skip the raw_evidence WebSearch leg, local Wiki+raw only - `/boss <topic> --refresh` — force EVOLUTION (even if the brand already has a report) - `/boss <topic> --no-judges` — skip Phase 4 (synthesis only) - `/boss <topic> --panel <name>` — use the specified panel.yaml - `/boss list` — list judged topics + version counts - `/boss panels` — list panel configs + judge identities IF the user asks "evaluate the X business carve-out plan" / "judge the X brand isolation strategy" / "have the 5 judges rate this SKU" / "run boss" / "run the verdict pipeline" / "5-judge deliberation" THEN invoke this skill. NOT WHEN: simple fact lookups (use `mcp__sage-wiki__query`) / single-judge opinions (use `*-perspective`) / pure technical research (use `research`) / everyday chitchat. boss is a heavyweight multi-judge deliberation — don't use it to answer 2-line questions.

source: local

This skill comes from the local ~/.claude/skills/boss/SKILL.md and is synced automatically by pnpm run sync:skills.

Description

Zero-dependency onboarding: install it and run /boss <topic> for a verdict deliberated independently by 5 judges — Phase 0 routing → 5-dimension parallel research → Lead synthesis → 1 anchor (baseline conviction) + 4 dimension judges scoring independently → panel_summary with anchor_delta + 30/90/365 attribution checkpoints.

Unlike the single-lens *-perspective skills, /boss is a full pipeline orchestration:

  • 1 anchor judge (anchor mental_models: trigger-threshold-action / step-skip detection / counter-party elimination)
  • 4 dim judges auto-selected per topic (panel auto-select: strategic/customer/organizational/product/brand/financial/cross_domain)
  • 5-lens scoring (reasoning / evidence / counter / falsifiability / real_world_resilience)
  • adversarial_view three fields enforced (CLAUDE.md §4.5)
  • Versioned freeze (chmod 444) + Wiki back-references

Router behavior is built in:

  • If reports/<brand-slug>/report.md already exists → EVOLUTION mode (diff plan + partial re-run)
  • Otherwise → FRESH mode (full Phase 1-5)

Trigger patterns:

  • /boss <topic or brand-slug> — standard invocation
  • /boss <topic> --quick — skip the raw_evidence WebSearch leg, local Wiki+raw only
  • /boss <topic> --refresh — force EVOLUTION (even if the brand already has a report)
  • /boss <topic> --no-judges — skip Phase 4 (synthesis only)
  • /boss <topic> --panel <name> — use the specified panel.yaml
  • /boss list — list judged topics + version counts
  • /boss panels — list panel configs + judge identities

IF the user asks “evaluate the X business carve-out plan” / “judge the X brand isolation strategy” / “have the 5 judges rate this SKU” / “run boss” / “run the verdict pipeline” / “5-judge deliberation” THEN invoke this skill.

NOT WHEN: simple fact lookups (use mcp__sage-wiki__query) / single-judge opinions (use *-perspective) / pure technical research (use research) / everyday chitchat. boss is a heavyweight multi-judge deliberation — don’t use it to answer 2-line questions.