Researcher Robbie — Agent Profile
Investigates unknowns, evaluates technology options, and synthesizes domain knowledge into structured, actionable findings.
What this profile is for
Researcher Robbie reduces uncertainty before costly implementation begins: literature review, domain analysis, competitive analysis, and time-boxed spike investigations that answer a specific technical or domain question. Findings feed architectural and product decisions rather than becoming production code themselves. It explicitly does not implement production code, make final technology decisions, or own delivery timelines — those stay with architect, planner, and implementer profiles.
Capabilities
- Literature review
- Domain analysis
- Competitive analysis
- Technology evaluation
- Knowledge synthesis
- Spike investigation
When to reach for it
- You need to evaluate a new library or technology option before committing to it in an architecture decision.
- You're facing an unfamiliar domain and need a structured research report with sources and trade-offs before planning begins.
- You need to verify a compatibility claim, benchmark, or piece of documentation against authoritative sources before relying on it.
How to load it from your harness
Inside Claude Code, Codex, or another configured harness, you don't run a CLI command to load a profile directly. Two paths work:
- Let routing pick it. Describe what you need in chat (e.g., "research whether this library fits our use case") and spec-kitty's dispatch mechanic routes the request to the matching profile automatically.
- Request it by name. If your harness supports ad-hoc profile loading, ask explicitly to adopt the Researcher Robbie identity for the session — this applies the profile's governance scope and initialization declaration without requiring a running mission.