agentic-research

Guides repository-first research using local code, official docs, and controlled external sources.

7|3|Updated Sep 23, 2025
One-click install
npx skills add https://github.com/islamu-ngo/Event --skill agentic-research-islamu-ngo
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: agentic-research
Source: https://github.com/islamu-ngo/Event/tree/main/.agents/skills/agentic-research
Command: npx skills add https://github.com/islamu-ngo/Event --skill agentic-research-islamu-ngo

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? It prevents agents from guessing API behavior, leaking sensitive data into external tools, or wasting effort on redundant searches by enforcing a strict evidence hierarchy: repository first, official documentation second, external research last. ## Core Features & Use Cases - Tiered Source Selection: Routes questions to local code, official docs (Context7, NuGet, Microsoft Learn), or external tools (Tavily, Exa) based on where the answer provably lives. - Security Boundaries: Defines what must never be sent externally (secrets, PII, tenant data, proprietary logic) and provides neutral query rewrite patterns. - Proportional Verification Matrix: Matches verification depth to change type, from docs-only edits to executable behavior changes requiring builds and architecture tests. - Use Case: When unsure how MudBlazor v9 handles dialog closing, inspect the repo first, then query official docs with a neutral question, and reconcile the answer against local code before applying it. ## Quick Start Ask the agent to research how a specific framework or library behaves, and it will check the repository first, then official documentation, then external sources while protecting sensitive context.

Frequently Asked Questions about agentic-research

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

FAQPage Schema
How do I research framework behavior without leaving the repository?▼

Inspect local code, tests, configuration, and .agents guidance first, since repo evidence outranks everything else. Only escalate to official docs like Context7, NuGet, or Microsoft Learn when the repository cannot answer the question.

When should I use external search tools versus official documentation?▼

Use official docs for package, framework, runtime, or migration uncertainty. Reserve external tools like Tavily or Exa for standards, RFCs, advisories, ecosystem comparisons, or when official docs are insufficient.

How do I ask external tools about my code without leaking secrets?▼

Reduce the question to a neutral description without tenant slugs, user IDs, internal URLs, or repo paths. Never send connection strings, tokens, PII, or large proprietary code blocks, and reconcile external answers against local code.

What verification is required for documentation-only changes?▼

Docs-only Markdown changes need structure review, link and path checks, and reference existence validation. Escalate to builds and architecture tests only when changes touch shared contracts, runtime behavior, or security-sensitive flows.

When should I stop researching and conclude an answer?▼

Stop when the repository already answers the question, official docs confirm the behavior, sources repeat the same conclusion, or two external sources agree. Do not keep escalating once evidence is sufficient.