Skip to main content

/tsh-research

info

Not invoked directly by users. To trigger research, use /tsh-implement — the Engineering Manager will automatically delegate to the Context Engineer when research is needed.

Agent: Context Engineer
File: .github/internal-prompts/tsh-research.prompt.md

Prepares a comprehensive context document for a task from a context engineering perspective.

How It's Triggered​

/tsh-implement <JIRA_ID or task description>

The Engineering Manager identifies that the task is missing necessary context and delegates research to the Context Engineer automatically.

What It Does​

  1. Gathers all information related to the task from the codebase, Jira, Confluence, and other sources.
  2. Analyzes the task thoroughly, including parent tasks and subtasks.
  3. Analyzes the tech stack, industry, and domain for best practices.
  4. Checks external links and Confluence pages linked to the task.
  5. Reviews Figma designs via Figma MCP (if linked to the task).
  6. Identifies ambiguities and asks for clarification before finalizing.
  7. Focuses on requirements, user stories, acceptance criteria, and key flows.

Skills Loaded​

  • tsh-task-analysing — Structured research process and output template.
  • tsh-codebase-analysing — Analyze existing codebase in the context of task requirements.

Output​

A .research.md file placed in specifications/<task-name>/:

specifications/
  user-authentication/
    user-authentication.research.md

The file contains all relevant information needed to build comprehensive context: task summary, requirements, user stories, acceptance criteria, assumptions, open questions, and suggested next steps.

tip

Review the generated research document carefully. Verify accuracy and iterate as many times as needed until the context is complete and correct.