deep-research

Conduct bounded, source-grounded research with cited evidence packs and decision recommendations.

Updated Jul 19, 2026

Search Tags:#research #source-grounded #citations #evidence #decision-making #evidence pack

Ready to run

npx skills add https://github.com/DagimAlemayehuu/AgenticEngineering --skill deep-research-dagimalemayehuu

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: deep-research
Source: https://github.com/DagimAlemayehuu/AgenticEngineering/tree/main/agentic-engineering/core/dependencies/research/deep-research
Command: npx skills add https://github.com/DagimAlemayehuu/AgenticEngineering --skill deep-research-dagimalemayehuu

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Research for engineering decisions often produces unbounded, uncited, or unverifiable findings. This Skill structures research into a decision-sized question with allowlisted sources, cited evidence, and a recorded recommendation so teams can act on traceable information. ## Core Features & Use Cases - Bounded Research Scope: Frames one decision-sized question and an allowlisted source plan before gathering evidence. - Cited Evidence Packs: Collects evidence with source URLs or durable local identifiers, including conflicting evidence and uncertainty. - Decision Output: Produces a recommendation, alternatives, and a decision owner, recorded through the Research phase gateway. - Use Case: When choosing between two database technologies for a project, use this Skill to gather cited comparisons, surface conflicting benchmarks, and record a redacted evidence pack with a clear recommendation. ## Quick Start Use the deep-research skill to investigate which caching strategy fits this project and produce a cited evidence pack with a recommendation.

Frequently Asked Questions about deep-research

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

FAQPage Schema

How do I run structured research for an engineering decision?▼

Frame one decision-sized question, define an allowlisted source plan, then gather cited evidence with URLs or durable local identifiers. The Skill produces a recommendation, alternatives, and a decision owner recorded through the Research phase gateway.

What does the deep-research skill output?▼

It outputs a decision-sized question, an allowlisted source plan, cited evidence with conflicting findings and uncertainty, a recommendation with alternatives and a decision owner, and a redacted evidence pack.

Does deep-research require external research tools or Obsidian?▼

No. The procedure is provider-neutral and does not assume a host-specific vault, Obsidian, or user filesystem. External research tools are optional adapters; if unavailable, it falls back to local research or reports the capability as unavailable.

Can deep-research write results to an external memory system?▼

No. It never silently writes to external memory systems and never invents citations. Research artifacts belong under the project's registered artifact paths and are recorded through the Research phase gateway.

When should I not use bounded deep research?▼

Avoid it for open-ended exploratory browsing without a decision at stake. The procedure requires a single decision-sized question and an allowlisted source plan, so unstructured discovery work falls outside its scope.