By name: research description: > Execute one focused research enquiry. Takes a specific question, searches for answers using available tools, produces structured findings. Used as a primitive by the Lisa loop and directly by humans for one-off research tasks. tools: - Read - Write - WebSearch - WebFetch - Bash compatibility: agents: all note: WebSearch and WebFetch availability varies by agent and environment.
/research
Execute a focused research enquiry. Not general-purpose search — a specific question with a specific output format.
Usage:
/research "What are the tradeoffs between JWT and session-based auth for a stateless API?"
/research --topic auth --enquiry 003 # part of a Lisa loop run
Output
Write findings to .loops/research/<topic-slug>/enquiry-NNN.md:
---
id: enquiry-001
topic: auth-strategy
question: "What are the tradeoffs between JWT and session-based auth for a stateless API?"
status: done
sources: []
---
## Findings
[Structured findings. Not a dump — curated. Answer the question directly, then support it.
Include: key facts, specific named sources where they matter, tradeoffs clearly stated,
what remains uncertain.]
## Key Points
- [Specific finding 1]
- [Specific finding 2]
- [What this means for the decision at hand]
## What This Doesn't Answer
[Be honest about gaps. What would you need to know to have higher confidence?]
Research Quality
Answer the question directly. Don't bury the answer in background. The findings should be usable by a planner or decision-maker reading them without further context.
Distinguish between:
- What you found from sources
- What you inferred
- What you're uncertain about
A shorter, honest findings document is more valuable than a long one with hedged claims.
After the Enquiry
Report: the question investigated, 2-3 key findings, what gaps remain. If this is part of a Lisa loop, the runner will handle synthesis and gap evaluation — just write the findings file.