Pathmode Skills for Claude Code

A skill pack that makes Claude Code aware of product intent — not just the code that implements it.

These skills wrap the @pathmode/mcp-server so Claude Code can build, stress-test, verify, and review intent specs through plain natural-language requests. No slash commands required.

Matt Pocock's skills make Claude Code a better engineer. This pack makes Claude Code build the right thing.

The pack

Skill	Use when
setup-pathmode-workflow	First-time project setup — capture test commands, issue tracker, status conventions
compile-intent	Build a structured spec for what to ship
verify-intent	Design the executable feedback loop for a spec (fastest check, manual fallback, shipped signal)
grill-intent	Stress-test an existing spec for weaknesses before code is written
split-intent-to-issues	Break a spec into paste-ready Linear / Jira / GitHub Issues tickets
review-against-intent	Check code changes against the intent's outcomes and constraints
handoff-intent	Capture decisions and discoveries at end of session

Install

# 1. Install the MCP server
npx @pathmode/mcp-server@latest setup

# 2. Install the skill pack
npx @pathmode/mcp-server install-skills            # project-local (.claude/skills/)
npx @pathmode/mcp-server install-skills --global   # global (~/.claude/skills/)

# 3. Restart Claude Code

That's it. Skills register at session start and auto-invoke when your request matches their description.

What auto-trigger feels like

You don't type a slash command. You just talk.

> help me write a spec for our checkout payment flow

[compile-intent skill auto-triggers]

  What's broken about it? When you say "checkout flow" — is the
  whole thing failing, or is there one specific moment where users
  bail?

> payment step times out and users get confused

  Proposed objective: "Reduce cart abandonment caused by payment
  step timeouts exceeding 3 seconds, where users see a spinner
  with no status feedback or recovery path."

  Does that capture it, or is there more to the pain?

Each question pairs with a best-guess answer drawn from the conversation. The user pushes back; the spec sharpens.

Why skills, not slash commands

Slash commands (/compile-intent) require you to know the command exists. If you've never read the docs, the command is invisible.

Skills carry a description field Claude Code reads at session start. When your natural-language request matches, the right skill runs automatically. You can use this pack effectively without reading any documentation past this README.

Modes

• Local mode (no API key) — Reads and writes intent.md in the project root. Free, no signup, fully offline.
• Team mode (PATHMODE_API_KEY set) — Syncs intents to a Pathmode workspace. Adds evidence, dependency graphs, constitution rules, and cross-agent context. Get a key from the Pathmode app.

Skills detect the mode at runtime and adapt — same skill names, more capability in team mode.

Format

Each skill is a SKILL.md file with YAML frontmatter and two sections:

---
name: skill-name
description: One paragraph Claude reads at session start to decide when to invoke this skill. The clearer this is, the more reliably it auto-triggers on the right requests.
---

<what-to-do>
The imperative the agent follows. Keep it short.
</what-to-do>

<supporting-info>
Examples, edge cases, rules — loaded when the skill triggers.
</supporting-info>

This matches the mattpocock/skills format exactly. If you've read his pack, you can read ours. The pattern composes — both packs can live in the same .claude/skills/ directory and reinforce each other.

What the skills wrap

Skills are the discovery layer. The engine is the @pathmode/mcp-server — 15+ tools, Socratic prompt, evidence handling, dependency graphs. Skills orchestrate calls to those tools.

You can also invoke the MCP prompts directly via slash commands if you prefer:

• /compile-intent — same as the compile-intent skill
• /implement-intent
• /review-risks
• /what-next

Skills wrap these in a more discoverable form.

Customizing

Each SKILL.md is just a markdown file. To customize behavior:

1. Edit the SKILL.md in .claude/skills/<skill-name>/
2. Restart Claude Code
3. The updated skill triggers next session

To add your own skills alongside these, drop a new <skill-name>/SKILL.md in the same directory. Follow the format above.

Related

• Pathmode — The intent engineering platform these skills connect to
• @pathmode/mcp-server — The MCP engine
• mattpocock/skills — The format precedent. Pairs naturally with this pack.

License

MIT — see LICENSE.