consultant（consultant 搜索）

Setup

If ~/consultant/ does not exist or is empty, initialize using setup.md and briefly inform the user that a local consulting workspace will be created.

When to Use

User needs structured consulting support: diagnosing issues, defining engagement scope, planning workstreams, and producing recommendations that can be executed.

Use this skill when unclear requests need framing, when stakeholders disagree, or when a decision memo, roadmap, or implementation plan is required.

Architecture

Working memory lives in ~/consultant/. See memory-template.md for the required structure.

~/consultant/
|-- memory.md                  # HOT: client context, preferences, active priorities
|-- engagements/               # One file per engagement
|   `-- YYYY-MM-client-topic.md
|-- decisions/                 # Decision logs with rationale and follow-up
|-- assets/                    # Reusable templates and frameworks
`-- archive/                   # Closed engagements and historical notes

Quick Reference

Core Rules

1. Diagnose Before Advising

Do not jump to solutions from surface symptoms.

Always establish:

• Objective: what decision or outcome the client needs
• Constraint set: time, budget, team capacity, policy limits
• Baseline: current state with evidence, not assumptions

Use discovery.md when context is incomplete.

2. Force Explicit Engagement Scope

Every consulting request must be translated into a clear contract of work.

Define in one block:

• Problem statement
• In-scope and out-of-scope boundaries
• Deliverables and acceptance criteria
• Timeline with review points
• Decision owners and approvers

If scope is fuzzy, state assumptions explicitly and mark them as risks.

3. Build Hypothesis-Driven Workstreams

Break work into workstreams that can be validated quickly.

For each workstream:

• Hypothesis: what must be true
• Evidence needed: data or stakeholder input
• Test method: interview, analysis, benchmark, pilot
• Decision trigger: what result changes the recommendation

Prefer fast tests that reduce uncertainty early.

4. Deliver Decision-Ready Outputs

Recommendations must be implementable, not abstract.

Every final recommendation includes:

• Why now: urgency and business impact
• Options considered and rejected
• Chosen option with tradeoffs
• Implementation sequence with owners
• Risks, mitigations, and fallback plan
• Leading metrics and review date

Use deliverables.md templates for consistency.

5. Manage Stakeholders Deliberately

Treat stakeholder alignment as a workstream, not a side task.

For key stakeholders, document:

• Position: sponsor, blocker, operator, approver
• Incentive: what they gain or lose
• Likely objection
• Engagement move: pre-wire, workshop, decision memo, escalation

Escalate early when decision rights are unclear.

6. Apply Quality and Risk Gates

Before sharing any recommendation, run the quality gate from quality-gates.md.

Minimum bar:

• Internal coherence (claims match evidence)
• Feasibility (capacity and sequencing are realistic)
• Financial sanity (benefit, cost, downside boundaries)
• Operational safety (no hidden critical dependency)

If a gate fails, revise before delivery.

7. Update Memory After Every Meaningful Interaction

Log new context in ~/consultant/memory.md and engagement files.

Persist only durable information:

• Preferred decision format
• Risk tolerance and time horizon
• Repeated constraints
• Confirmed stakeholder map changes

Do not store secrets, credentials, or unrelated personal data.

Engagement Flow

Use this sequence for new consulting requests:

When requests are urgent, run a compressed version but keep all five stages explicit.

Common Traps

• Over-scoping the engagement -> work stalls and trust drops
• Presenting recommendations without options -> stakeholders feel forced and resist
• Ignoring decision rights -> quality work gets blocked in governance
• Delivering analysis without action sequence -> no execution despite agreement
• Hiding assumptions -> recommendation fails when assumptions break
• Treating dissent as noise -> critical implementation risks remain invisible
• Confusing activity with impact -> many tasks, no measurable result

Scope

This skill covers:

• Consulting discovery and scoping
• Problem diagnosis and structured analysis
• Decision memo and roadmap creation
• Stakeholder alignment planning
• Quality and risk gating before delivery

Use complementary skills for deep specialty work:

• Finance-heavy modeling -> cfo
• Executive leadership dynamics -> ceo
• Competitive positioning deep dives -> strategy
• Pricing architecture and packaging -> pricing

Security & Privacy

Data that may leave your machine:

• Only what the user explicitly asks to include in external tools during normal agent operation

Data that stays local:

• Context and engagement notes in ~/consultant/

This skill does NOT:

• Access undeclared external endpoints by itself
• Read files outside consulting context without user need
• Store secrets or credentials in memory files

Related Skills

Install with clawhub install <slug> if user confirms:

• business - Validate initiatives and prioritize strategic decisions
• strategy - Build competitive positioning and strategic option maps
• ceo - Support executive-level decision framing and communication
• cfo - Model financial impact and downside scenarios
• pricing - Design pricing structures and packaging decisions

Feedback

• If useful: clawhub star consultant
• Stay updated: clawhub sync