From product-shaping
Invoke this skill when the user asks how to USE the product-shaping plugin day-to-day — which /shape-* command to run next, where artifacts land, how to switch between greenfield and brownfield workflows, troubleshooting errors, or working on a specific OS (Windows, Linux, macOS). Covers all /shape-* slash commands, artifact locations, pipeline stages, common errors, and platform-specific tips. Excludes: first-time installation (use shape-cli-setup instead), developing or contributing to the plugin source code, and general programming help.
How this skill is triggered — by the user, by Claude, or both
Slash command
/product-shaping:shape-cli-guideThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
This skill helps users work with the `product-shaping` plugin after it is already installed. If the user has not installed the plugin yet, hand off to the **shape-cli-setup** skill instead.
This skill helps users work with the product-shaping plugin after it is already installed. If the user has not installed the plugin yet, hand off to the shape-cli-setup skill instead.
Before giving any guidance, gather context silently — run these checks and remember the results. Do not print raw output to the user.
echo "$OSTYPE" 2>/dev/null || echo "win32"
Use the result to tailor path separators, shell syntax, and clipboard commands throughout your answers:
| OS | Shell | Home var | Clipboard | Temp dir |
|---|---|---|---|---|
macOS (darwin*) | zsh / bash | $HOME | pbcopy | $TMPDIR |
Linux (linux-gnu*) | bash / zsh | $HOME | xclip -selection clipboard or xsel --clipboard | /tmp |
Windows (win32 / MSYS / Git Bash) | PowerShell / cmd | %USERPROFILE% | clip.exe | %TEMP% |
Check which skills directory exists to infer the active AI tool:
ls -d ~/.claude/skills/shape-idea 2>/dev/null && echo "claude-code"
ls -d ~/.cursor/skills/shape-idea 2>/dev/null && echo "cursor"
ls -d ~/.github/skills/shape-idea 2>/dev/null && echo "copilot"
ls -d ~/.agents/skills/shape-idea 2>/dev/null && echo "codex"
ls -d ~/.ai/skills/shape-idea 2>/dev/null && echo "generic"
ls -d ~/.claude/skills/shape-idea 2>/dev/null || ls -d ~/.cursor/skills/shape-idea 2>/dev/null || echo "NOT_FOUND"
If the plugin is not found, redirect to shape-cli-setup.
Use the environment context from Step 1 to personalize every answer. Always use the OS-appropriate shell syntax, paths, and commands. Never show macOS-specific commands to a Windows user or vice versa.
/shape-workflow-guide — Where do I start?The recommended first command. It asks 1-2 orienting questions (greenfield vs brownfield, idea vs existing codebase) and names the exact next /shape-* command to run.
/shape-idea → context/foundation/shape-notes.md
↓
/shape-prd → context/foundation/prd.md
↓
/shape-tech-stack → context/foundation/tech-stack.md
↓
/shape-bootstrap → scaffolded project in cwd
↓
/shape-agents-md → AGENTS.md in cwd
/shape-idea → context/foundation/shape-notes.md
↓
/shape-prd → context/foundation/prd.md
↓
/shape-stack-assess → context/foundation/stack-assessment.md
↓
/shape-health-check → context/foundation/health-check.md
↓
/shape-agents-md → AGENTS.md in cwd
/shape-roadmap → context/foundation/roadmap.md
↓
/shape-new <id> → context/changes/<change-id>/
↓
/shape-frame → context/changes/<change-id>/design/frame.md
↓
/shape-research → context/changes/<change-id>/research/
↓
/shape-plan → context/changes/<change-id>/plan.md
↓
/shape-plan-review → review report
↓
/shape-implement → implemented code + updated plan.md Progress
↓
/shape-impl-review → review report
↓
/shape-archive → context/archive/<date>-<change-id>/
/shape-infra-research → deployment platform research
/shape-rule-review → audit existing AI rules
/shape-lesson → capture a learning or pattern
/shape-init → scaffold context/ directory explicitly
All product-shaping artifacts are written under context/ in the project's main repo root (worktree-aware). The exact paths:
| Skill | Output path | Type |
|---|---|---|
/shape-idea | context/foundation/shape-notes.md | Markdown notes |
/shape-prd | context/foundation/prd.md | Markdown PRD |
/shape-tech-stack | context/foundation/tech-stack.md | Markdown spec |
/shape-stack-assess | context/foundation/stack-assessment.md | Markdown report |
/shape-health-check | context/foundation/health-check.md | Markdown report |
/shape-agents-md | AGENTS.md (in cwd) | Markdown instructions |
/shape-roadmap | context/foundation/roadmap.md | Markdown plan |
/shape-new | context/changes/<change-id>/change.md | Markdown tracking |
/shape-frame | context/changes/<change-id>/design/frame.md | Markdown brief |
/shape-research | context/changes/<change-id>/research/ | Markdown + data |
/shape-plan | context/changes/<change-id>/plan.md | Markdown plan |
/shape-plan-review | conversation output only | Review report |
/shape-implement | code in cwd + updated plan.md | Source files |
/shape-impl-review | conversation output only | Review report |
/shape-archive | context/archive/<date>-<change-id>/ | Archived folder |
None of these are committed to git by default. context/ is excluded via git info/exclude automatically.
To switch from greenfield to brownfield (or vice versa), simply run /shape-workflow-guide and answer the orienting questions differently. The pipeline adapts automatically — there is no config file to edit.
To switch AI tools (e.g., from Claude Code to Cursor):
product-shaping/skills/ directory into the new tool's skills folder.context/ stays in the same repo root regardless of tool — no migration needed.clip.exe on Windows. If a skill outputs a clipboard command, it will fall back silently if clip.exe is unavailable.context/changes/my-change/) work fine on Windows in most modern shells.xclip -selection clipboard or xsel --clipboard. If neither is installed, clipboard operations fail silently. Install with sudo apt install xclip (Debian/Ubuntu) or sudo dnf install xclip (Fedora).~/.config/ or $XDG_CONFIG_HOME if set.pbcopy — works out of the box.~/.claude/ (or tool-specific equivalent).When the user reports a problem, follow this sequence:
# Check if context/ exists and where it is anchored
if git rev-parse --is-inside-work-tree >/dev/null 2>&1; then
GIT_COMMON_DIR="$(git rev-parse --git-common-dir)"
CONTEXT_ROOT="$(dirname "$(cd "$GIT_COMMON_DIR" && pwd)")"
echo "Context root: $CONTEXT_ROOT"
ls -la "$CONTEXT_ROOT/context/" 2>/dev/null || echo "context/ not found"
else
echo "Not inside a git repo — context/ would be at $(pwd)/context/"
fi
This catches the most common issues: wrong directory, missing context/, or plugin not installed.
| Symptom | Likely cause | Fix |
|---|---|---|
| "context/ not found" | First run — context not initialized | Run /shape-init or any /shape-* command; it self-bootstraps |
git status shows context/ files | Git exclude not applied | Check $(git rev-parse --git-common-dir)/info/exclude for context/ line |
context/ missing from worktree B but present in A | Context anchored to wrong repo | Ensure both worktrees belong to the same repo; run diagnostics above |
/shape-* command not found | Plugin not installed for this AI tool | Run /shape-cli-setup or copy skills to the correct directory |
permission denied writing files | Directory not writable | Check directory permissions; on POSIX: chmod u+w <dir> |
If the context directory is corrupted or you want a clean slate:
On macOS/Linux:
rm -rf "$(git rev-parse --show-toplevel)/context"
# Then run /shape-init to re-scaffold
On Windows (PowerShell):
Remove-Item -Recurse -Force "$(git rev-parse --show-toplevel)\context"
# Then run /shape-init to re-scaffold
This clears all shaping notes but leaves source code untouched. The next /shape-init recreates the skeleton.
pbcopy to a Windows user. Never show %APPDATA% to a macOS user.https://raw.githubusercontent.com/patrykkopycinski/patryks-treadmill-claude-plugins/refs/heads/main/plugins/product-shaping/README.md/shape-* command writes to the wrong place, it is a plugin setup question. If the user's code doesn't compile after /shape-bootstrap, it is a project question.npx claudepluginhub patrykkopycinski/patryks-treadmill-claude-plugins --plugin product-shapingCreates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.