Skills for the whole product R&D process: strategy, UXR, PM, design, QA, launch. For Claude Code, Hermes Agent, Codex, and other skill-aware agents.
Overview
Great product work isn't one skill, it's a pipeline: decide what to build and why, understand the user, spec it, design it, prove it works, then ship and price it. Most teams are strong at one or two of these stages and improvise the rest. This is a toolkit that gives a skill-aware agent a real method for each stage, derived from skills used in day-to-day product building. These are agent skills that work with Claude Code, Hermes Agent, Codex, and other skill-aware agents. Each stage holds two focused skills. thinking is the cross-cutting pair you reach for at any stage when you're stuck, deciding, or about to commit. Point any skill-aware agent at this repo and ask it to load the skill, or copy a skill folder into your agent's skills directory. Each skill is self-contained: a SKILL.md entry point plus references/ and scripts/ it loads as needed.
README
pm-skills
Skills for the whole product R&D process, from strategy to launch.
Great product work isn’t one skill, it’s a pipeline: decide what to build and why, understand the user, spec it, design it, prove it works, then ship and price it. Most teams are strong at one or two of these stages and improvise the rest. This is a toolkit that gives a skill-aware agent a real method for each stage, derived from skills used in day-to-day product building.
These are agent skills that work with Claude Code, Hermes Agent, Codex, and other skill-aware agents.
The pipeline
STRATEGY -> UXR -> PM -> DESIGN -> QA -> LAUNCH
(thinking runs underneath every stage)
Each stage holds two focused skills. thinking is the cross-cutting pair you reach for at any stage when you’re stuck, deciding, or about to commit.
| Category | Skills | Description |
|---|---|---|
| design | 2 | Make it good. Review a design against usability heuristics with before/after fixes, then iterate on mockups from critique with a keep/change/kill rationale so each version is measurably better than the last. |
| launch | 2 | Ship it and price it. Plan the launch (audience, channels, sequencing, messaging) and set pricing and packaging so the thing you built actually converts and captures value. |
| pm | 2 | Spec and justify the thing. Make the case for a feature as a business bet tied to a funnel metric, and diagram the system so the team shares one picture of how it works. |
| qa | 3 | Prove it works. Verify a change actually runs at its real surface with runtime evidence instead of trusting tests, dogfood the product as a difficult user, and attack it as a hostile persona to find the issues before customers do. |
| strategy | 2 | Decide what to build and why. Diagnose the real obstacle, choose a guiding policy, and pressure-test whether a plan is actually a strategy or just goals plus fluff. Size up competitors before you commit. |
| thinking | 2 | Cross-cutting reasoning used at every stage. Generate options fast with surfaced assumptions, and stress-test a decision from multiple perspectives before you commit to it. |
| uxr | 2 | Understand the user before you build. Run synthetic user studies to test designs against realistic personas, and map the end-to-end journey to find the moments that decide whether a user stays or leaves. |
The skills
| Stage | Skill | What it does |
|---|---|---|
| strategy | strategy-kernel | Diagnose the real obstacle and build (or debunk) a strategy with Rumelt’s kernel: diagnosis, guiding policy, coherent action. Catches goals-dressed-as-strategy. |
| strategy | competitive-analysis | Map the field with battlecards grounded in real evidence, not vibes. |
| uxr | synthetic-userstudies | Run AI personas through a design to surface reactions and objections early. |
| uxr | journey-mapping | Chart the stages, emotions, and pain points across a flow, and storyboard the moments that decide whether a user stays. |
| pm | pitch-creator | Make the case for a change as a funnel-impact bet: working prototype, expected metric deltas, risks, alternatives, recommendation. |
| pm | architecture-diagram | Render the system as a clean SVG so the team shares one picture. |
| design | design-review | Structured heuristic evaluation with concrete before/after fixes. |
| design | design-iteration | Turn critique into the next version with a per-change rationale trail. |
| qa | verify-feature | Prove a change runs at its real surface with runtime evidence, not passing tests. |
| qa | dogfood | Exercise the product as a demanding user; file reproducible issues with evidence. |
| qa | adversarial-ux-test | Attack the product as a hostile or low-patience persona to surface the failures a cooperative user never hits. |
| launch | launch-strategy | Plan the release: audience, channels, sequencing, messaging. |
| launch | pricing-strategy | Set price, packaging, and paywall structure from willingness-to-pay and unit economics. |
| thinking | quick-brainstorm | Several distinct options with rough probabilities and the assumptions each rests on. |
| thinking | another-perspective | Multi-perspective council to red-team a plan and synthesize the tensions. |
Using a skill
Point any skill-aware agent at this repo and ask it to load the skill, or copy a skill folder into your agent’s skills directory. Each skill is self-contained: a SKILL.md entry point plus references/ and scripts/ it loads as needed.
git clone https://github.com/exiao/pm-skills
# then, in your agent:
# "use strategy-kernel to pressure-test our Q3 strategy doc"
# "map the onboarding journey and storyboard the drop-off moment"
# "verify this feature actually works before I merge"
License
Apache 2.0. See LICENSE.txt.
Recommended Tools
Try a different keyword or remove a filter.
Install
npx skillfish add exiao/pm-skills