Research memo
Use for claims, evidence, citations, risks, and periodic agent refreshes.
Copy these templates when starting a new artifact. Each one demonstrates the intended shape of a document that humans can review, agents can patch, and CI can render.
Use for claims, evidence, citations, risks, and periodic agent refreshes.
Use for product or architecture decisions with options, accepted status, and follow-up tasks.
Use for API designs, implementation plans, diagrams, validation rules, and rollout notes.
Use when an agent should regularly check stale sources and update specific target blocks.
# Research Memo
::summary
One paragraph on the question, answer, and confidence.
::
::claim{id="core-claim" confidence=0.72}
The central claim goes here.
::
::evidence{for="core-claim" source="source-id"}
Evidence that supports the claim.
::
# Decision Record
::decision{id="decision" status="proposed" owner="team"}
We propose option B because it maximizes learning while keeping rollback cheap.
::
::agent_task{id="decision-follow-up" scope="post-decision"}
After two weeks, verify whether the decision criteria still hold.
::
# Technical Spec
::summary
What is being built, who it serves, and what is explicitly out of scope.
::
::diagram{kind="mermaid"}
flowchart LR
Source --> Parser --> AST --> Renderer
::
# Agent Refresh Pack
::agent_task{id="refresh-task" scope="weekly"}
Check citations, update stale evidence, and leave unrelated blocks byte-identical.
::
id.status, owner, confidence, and source attributes instead of burying metadata in prose.agent_task when the document is expected to be maintained over time.noma check before publishing the rendered artifact.