‹ Engineering recipesEngineering & SRE · Agent Blueprint
New-engineer onboarding
Who owns this service and what should I know? Owner, on-call rotation, last three incidents, recent design decisions, all in one card.
Side by side: token usage, with and without Paperbase
~68k tokens saved · 68% less work for the agentWithout Paperbase
~100k tokens
- Search the wiki for the service page and skim it.
- Guess who owns the service by reading old tickets.
- Grep incident archives for the service name across the last year.
- Read design docs one at a time to infer recent decisions.
- Ask three different people for the same answers.
- Stitch a picture together from fragments no one owns.
With Paperbase
~32k tokens
- Pull the Service entity: owner, on-call, dependencies.
- Pull the last three incidents attached to it, summarized.
- Pull recent Design Decisions about it, with rationale.
- Compose the one-card briefing from typed records.
Agent prompt
You are onboarding a new engineer onto {{service_name}}.
Using Paperbase:
1. Pull the service entity: owner, on-call rotation, dependencies, related repositories.
2. Pull the last three incidents for the service, summarized in two lines each.
3. Pull the last 90 days of Decisions touching the service, with rationale and who made them.
4. Pull open Commitments or TODOs on the service.
Output a single briefing card:
- Who owns it, who is on call, what it depends on.
- Last three incidents and what changed because of them.
- Recent design decisions, one line each.
- Open work with owners.Placeholders in {double_braces} are inputs the agent will ask for at runtime. Give the prompt to any agent connected to Paperbase, and the rest grounds in your own memory graph.