ask

Answers codebase and web research questions through parallel researcher agents and cited synthesis.

2|Updated Apr 4, 2026
One-click install
npx skills add https://github.com/DoctorMozg/claude-pipelines --skill ask-doctormozg
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ask
Source: https://github.com/DoctorMozg/claude-pipelines/tree/main/plugins/mz-dev-pipe/skills/ask
Command: npx skills add https://github.com/DoctorMozg/claude-pipelines --skill ask-doctormozg

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Getting a trustworthy answer about how a codebase works, why a technology was chosen, or whether a project supports a capability usually requires digging through files and documentation manually. This Skill automates that read-only research pass and returns a direct, source-cited answer without modifying any code. ## Core Features & Use Cases - Question Classification: Routes each question as codebase, external, or hybrid so research effort goes to the right sources. - Parallel Research Agents: Dispatches up to 3 codebase researcher agents and 2 web researcher agents in parallel, with retry and partial-finding handling. - Cited Synthesis & Report: Produces a direct answer with file:line and official-source citations, saved as a report under .mz/reports/. - Use Case: Ask "how does the auth middleware refresh tokens" and receive a grounded answer citing the exact files and lines, plus a persistent markdown report. ## Quick Start Ask the assistant to use the ask skill to answer how the WebSocket reconnection backoff works in this repository.

Frequently Asked Questions about ask

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I get an answer about how code works in a repository?

Ask a natural-language question like "how does the auth middleware refresh tokens". The skill decomposes it into sub-questions, dispatches parallel codebase researcher agents, and returns a direct answer with file:line citations plus a saved report.

How does codebase question answering handle external library questions?

Questions are classified as codebase, external, or hybrid. External and hybrid questions trigger web research agents that consult official, version-matched sources, grounded in the stack and dependency versions detected from the repository.

Does the ask skill modify any source code?

No, the pipeline is strictly read-only. It only writes working artifacts under .mz/task/ and a final markdown report under .mz/reports/, never touching source files.

When should I not use a question-answering research skill?

Do not use it when you want code modified or built, a diagram-rich walkthrough, or a hypothesis proven with tests. It is also unnecessary for one-line questions answerable directly from context without file or web lookups.

What happens when research agents return empty or conflicting findings?

Empty or errored agents are re-dispatched once, then the pipeline proceeds with partial findings and notes the gap. Conflicting findings are surfaced in the answer with a labelled confidence level rather than being hidden.