Research codebase for facts
2_researchcommandsetup L1★38
matanshavit/qrspi ↗What it does
Conduct objective codebase research answering specific technical questions
Best for
Objective technical investigations in large codebases requiring evidence-based answers to specific architectural or behavioral questions.
Inputs
- · Question directory path (thoughts/qrspi/<id>/)
- · Codebase context
- · Existing question files
Outputs
- · Facts-only research findings (no opinions)
- · Code examples and evidence
- · Answers keyed to original questions
Requires
- · Codebase grep/search
- · Code reading (Read tool)
- · Claude Opus model (specified)
Preconditions
- · Question directory with pre-specified questions
- · Codebase readable and fully available
- · Project structure known
Failure modes
- · Question unanswerable → documents limitation clearly
- · Code missing or private → reports inaccessibility
- · Opinion leakage → enforces facts-only format
Trust signals
- · Model specified as Opus (high-capability reasoning)
- · Facts-only constraint prevents opinion injection
- · Evidence pulled directly from codebase (not inference)
- · Question-keyed format maintains traceability