Stats
Actions
Available In
Tags
Plugin
ideation
Structured ideation: generate high-impact feature ideas and run synthetic user-research simulations to decide what to build next.
What's Inside
Skills2
fan-out-user-simulation
/fan-out-user-simulation
Fans out a diverse cohort of simulated user personas 'using' the product (with mocked product responses) in parallel sub-agents, then synthesises their traces into ranked feature recommendations. Use when 'simulate users', 'user simulation', 'synthetic user research', 'what would users want', or deciding what to build next. Not for persona-free feature brainstorming (use generate-10-ideas).
generate-10-ideas
/generate-10-ideas
Generate 10 brilliant ideas for powerful new functionality. Use when brainstorming features, improvements, or innovations for a system. Internally generates 100 candidates and filters to the top 10 most impactful, pragmatic, and innovative ideas.
README
Skills
A collection of reusable agent skills that implement the Portable Task Workflow -- an AI-agent-driven feature delivery lifecycle from requirements through merged PR.
requirements.md -> solution.md -> context.md -> plan.md -> worktree execution -> PR -> review -> merge -> feedback
Planning happens on main. Implementation happens on feature branches in isolated git worktrees. Each stage is a discrete skill invoked via slash command.
Available Skills
| Skill | Description |
|---|---|
new-task | Create a new task folder with requirements.md from user input |
new-solution | Write a solution design (solution.md) by exploring approaches and tradeoffs |
new-plan | Write context.md and plan.md by gathering codebase context and breaking work into phases |
execute-task | Autonomously implement a planned task end-to-end using worktree isolation |
delegate-task | Dispatch task execution to an idle Claude Code pane in a tmux session |
executor-status-check | Monitor all executor panes in a tmux session and report status |
commit-task | Verify completeness of planning docs and commit them to main |
complete-task | Finalize a feature branch and merge it to main with PR, testing, and cleanup |
code-review | Perform a structured code review with severity labels |
review-task | Review task planning documents and leave structured inline comments |
request-codex-review | Send task planning docs to Codex for review, then resolve any comments |
address-feedback | Work through open PR review threads by fixing code and resolving threads |
resolve-comments | Resolve inline review comment threads in planning docs (markdown and HTML) |
task-feedback-analyser | Extract recurring patterns from completed task feedback into workflow rules |
generate-10-ideas | Brainstorm 10 ideas by generating 100 candidates and filtering to the top 10 |
revise-readme | Update README and documentation to reflect the current state of the project |
Installation
Install all skills (from remote)
npx skills install mistakenot/skills -s '*' -a claude-code codex -y
Install a single skill (from remote)
npx skills install mistakenot/skills -s new-task -a claude-code codex -y
This installs into .claude/skills/ and .agents/skills/.
Development
Files in ./skills/ are compiled output. Edit the source files in ./src/ and run the Makefile targets:
make compile # Compile src/ -> skills/
make install # Compile + install locally into .claude/skills/ and .agents/skills/
make lint # Lint all skills with autoskill
make check # Compile + lint (pre-commit check)
Stats
Version0.2.0
Released
LanguageJavaScript
Stars0
MaintenanceGood
LicenseMIT
Last Commit
Added
Available In
Tags
Categories
