Skip to content
Back to skills

Ask User

ASecurity

Ask which installed skill or workflow fits the user's situation.

  • 5 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added October 21, 2026
developmentrailscode-reviewgitperformancedocumentation

Works with

  • cli

Security analysis

A100/100

Pro scans all 3 files and shows the line behind each finding

Scanned October 21, 2026

npx -y skills add collectiveai-team/botica --skill ask-user --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Ask User?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Ask User
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/collectiveai-team-ask-user/badge)](https://www.skillsdirectory.com/skills/collectiveai-team-ask-user)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: ask-user
description: Ask which installed skill or workflow fits the user's situation.
disable-model-invocation: true
---

# Ask User

Route the user's situation to the smallest suitable workflow. Explain the recommendation briefly, then ask whether they want to invoke it. Do not invoke another user-invoked skill without confirmation.

Before stating what a skill does or recommending a step be skipped, read that skill's `SKILL.md`: the summaries here are for orientation only.

## Main Flow

For an idea that needs to ship:

1. Use `/grill-with-docs` to sharpen an idea in a codebase while updating `GLOSSARY.md` and ADRs. Use `/grill-me` when there is no codebase or durable project documentation is not wanted.
2. If a design question needs a runnable answer, use `/handoff` to branch into a fresh `/prototype` session, then `/handoff` the findings back.
3. For work that fits one session, use `/implement` directly. For multi-session work, use `/to-spec`, then `/to-tickets`, and start a fresh `/implement` session for each unblocked ticket. Use `/implement-spec` instead when the whole ticket graph should run in parallel worktrees and land on one integration branch.

`/implement` drives `/tdd` at agreed seams and closes with `/code-review`. Use `/tdd` or `/code-review` directly when only that focused workflow is needed. Before creating, updating, or marking a PR ready, `/engineering-pr-review` runs the gate on top of that (`botica lint`, `/code-review`, the checks of the diff) and reports what it ran. `/test-smell-review` applies whenever tests are written or edited; reach for it directly to audit an existing suite for tests that pass without protecting anything.

Keep grilling, specification, and ticketing in one context window when practical, so each step builds on the same thinking. Between independently implementable tickets, clear context; see Phase Boundaries.

## On-Ramps

- Incoming bugs and requests that are not already agent-ready: `/triage`.
- A hard bug, intermittent failure, or performance regression: `/diagnosing-bugs`.
- A huge, foggy effort whose path cannot fit one session: `/wayfinder`, then `/to-spec` when the decision map is clear.
- A completed coding session: `/retro` to identify improvements to the agent's environment and guardrails.
- A PR body to write: the internal `/pr` reads the target repo's `.github/pull_request_template.md`, after the `/engineering-pr-review` house gate.

Do not triage tickets produced by `/to-tickets`; they are already agent-ready. Do not send a resolved wayfinder map directly to `/implement` unless the effort proved genuinely small.

## Codebase Health

- Survey a codebase for deepening opportunities: `/improve-codebase-architecture`.
- Design a chosen module or seam using deep-module vocabulary: `/codebase-design`.
- Refactor an existing repo in tested slices, keeping its local conventions, with botica for the rules: `/engineering-refactor`.
- Resolve unclear or overloaded domain terminology: `/domain-modeling`.

## Phase Boundaries

A phase is a chunk of work inside a session: the grilling, the implementation, the review. Decide what to do with the context only at the boundary between two phases; mid-phase, continue or split the remaining work into subagents. Ask these in order and take the first yes.

1. **Continue** when the next phase needs this one as a primary source, or the remaining context window comfortably fits it. Continue costs nothing and loses nothing, so rule it out first.
2. **Clear** when everything in this session is disposable to what comes next.
3. **Handoff** (`/handoff`) when something has to travel: a different harness, a different directory or worktree, another person, or a side task forked mid-phase. What it buys is portability, so skip it when nothing travels.
4. **Subagent** when the task is scoped tightly enough to run unattended and report back.
5. **Compact** otherwise. This is the default landing spot, not the first reach; pass it an instruction so the summary keeps what the next phase needs.

Every option except Continue replaces the session with a summary of it, which is why Continue is questioned first.

## Standalone

- `/prototype`: answer one state, logic, or UI design question with throwaway code.
- `/research`: delegate primary-source research and capture cited findings in the repo.
- `/grilling`: run the interview primitive directly, with no wrapper flow around it.
- `/teach`: learn a concept over multiple sessions in a stateful workspace.
- `/writing-for-agents`: guide the writing of documents agents consume: skills, `AGENTS.md`, pointed-at docs.
- `/to-questionnaire`: turn a decision or interview into a shareable questionnaire.
- `/wizard`: turn a procedure only a human can perform (provisioning, credentials, CI secrets, clicking through a third-party dashboard, a one-off migration) into an interactive script that walks them through it. Not for steps an agent can perform itself.
- `/wait-what`: re-pitch the last message when it did not land. Usable mid-conversation inside any other skill.
- `/journalist`: record or search session notes under `.journals/`.

## Precondition

Use `/setup-matt-pocock-skills` once per repository before engineering flows that depend on its issue tracker, labels, or domain documentation layout.

Credits and sources: [`references/CREDITS.md`](references/CREDITS.md).

Files in this skill

  • SKILL.md5.3 KB
  • agents/openai.yaml137 B
  • references/CREDITS.md987 B

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.