tools-only Product Manager: Typical Capabilities
Which capabilities from companion-kits/public-kits/capabilities/ this persona typically uses, and how.
Core Capabilities (Always Include)
Claude Code Configuration
Role: The infrastructure layer that makes all other capabilities work reliably. How it's used: Distributes instructions across the reliability hierarchy — hooks for enforcement, skills for workflows (LOW freedom for generative tasks like email drafting and report generation), agents with preloaded skills for isolated execution, rules for domain patterns, lean CLAUDE.md for identity. PM companions are the heaviest users because generative tasks (emails, prep reports, strategic recommendations) need maximum enforcement to prevent training-data pattern substitution. Customization: SessionStart hook includes skill-checking reminder. Communication standards enforced via hooks or low-freedom skills. All multi-phase commands (process-meeting, write-followup, start-day) defined as low-freedom skills with compliance checkpoints.
Reverse Prompting
Role: The foundational interaction pattern for PM companions. How it's used: Seeding commands (explore, process) are predominantly reverse prompting — Claude asks, founder answers. The PM extracts and structures what the founder knows but hasn't articulated. Customization: PM voice behaviors shape the reverse prompting style (Skeptic, Connector, Narrower, Tester, Shaper).
Context Ecosystem
Role: The standard context capture and persistence layer. How it's used: requirements.md, constraints.md, decisions.md, questions.md — plus PM-specific product context files (vision.md, user-segments.md, competitive-landscape.md, feature-concepts.md). Customization: Product context files extend the base ecosystem with PM-specific artifacts.
Strategic Planning
Role: Hypothesis-driven decision making as the central methodology.
How it's used: The product hypothesis serves as the strategy anchor. Every command references it. /hypothesis manages it. /spec and /plan use it as the decision lens for scope and priority.
Customization: The hypothesis format, confidence levels, and test plan structure are PM-specific.
Common Capabilities (Include When Relevant)
Meeting Processing
When to include: When the founder has regular meetings, customer conversations, or stakeholder discussions that produce insights.
How it's used: /process transcript ingests meeting transcripts (Granola, etc.) and extracts product insights, connecting new information to existing product context and hypothesis.
Insight Feedback Loop
When to include: When the PM project runs for more than a few sessions and insights need to compound.
How it's used: Observations from each session feed into insights-log.md. /synthesize finds patterns across sessions. The PM voice references accumulated insights when challenging new ideas.
Session Hygiene
When to include: When session continuity matters and the working rhythm is established.
How it's used: /status at session start (orientation), /checkpoint at session end (capture). Ensures nothing important stays only in conversation context.
Knowledge Zones
When to include: When the PM project accumulates enough material that navigation becomes important. How it's used: Organized reference (product/), field knowledge (tracking/), insights (tracking/insights-log.md), session history (tracking/session-log.md), decisions (context/decisions.md).
Rarely Used Capabilities
Mentor Framework
When to include: Only if the PM project involves a specific domain where named mentor personas add value (rare for pure PM work).
Craft Assessment
When to include: Not typically used for PM projects.
Process Evolution
When to include: When the PM project has been running long enough that commands need tuning. Uses the /record + /evolve pattern.