‹ Engineering recipesEngineering & SRE · Agent Blueprint
Retro prep
Sprint blockers and how they were resolved. Every blocker mentioned in standup or thread, and the resolution path.
Side by side: token usage, with and without Paperbase
~107k tokens saved · 82% less work for the agentWithout Paperbase
~130k tokens
- Read every standup note from the sprint.
- Skim Slack threads for mentions of blockers and workarounds.
- Check the board for stuck cards and re-opened tickets.
- Read the previous retro notes for themes that recurred.
- Reconstruct resolution paths from fragmentary messages.
With Paperbase
~23k tokens
- Pull the sprint's Interactions - standups, threads, review meetings.
- Ask Paperbase to extract Blockers and their owners.
- For each blocker, trace the resolution path through the record.
- Spot blockers that recurred across sprints automatically.
Agent prompt
You are preparing the sprint retro for the sprint ending {{date}}.
Using Paperbase:
1. Pull Interactions from the sprint: standups, Slack, review meetings.
2. Extract every Blocker mentioned, with who owned it.
3. For each blocker, trace how it was resolved - or whether it is still open.
4. Compare against prior sprints for blockers that keep recurring.
Output:
- Blockers, ranked by how much of the sprint they consumed.
- For each: owner, resolution path (or still open), and the moment it was first raised.
- Recurring blockers from prior sprints, called out.
- Three questions for the retro.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.