summarizer — Claude Code prompt: what it does, how to install
The summarizer Claude Code prompt condenses an infrastructure optimization search into a factual handoff. It keeps candidate results and unresolved hypotheses for later turns.
no license ★ 99 updated 2026-11-12
.claude/commands/summarizer.mdWhat summarizer does
summarizer turns earlier discussion of infrastructure optimization into a compact record for the next steps. It keeps candidate IDs and changes, screening outcomes with reasons, StepTime and peak memory, the current best result, failures and their causes, plus hypotheses still under review.
It removes repetition and speculation, retaining only details present in the conversation. The reply must fit within limit_tokens and contain exactly one JSON object with a summary key. In Claude Code, install it as .claude/commands/summarizer.md.
What it helps with
- condense earlier turns in an infrastructure optimization search
- carry screening outcomes and rejection reasons into later turns
- preserve the current best result and unresolved hypotheses
Example prompts for the agent
- Summarize the earlier infrastructure optimization search so we can continue.
- Record candidate changes, screening results, StepTime, and peak memory.
- Return the current best, failure causes, and open hypotheses as JSON.
How to install summarizer
Manually: put the files in place
.claude/commands/summarizer.mdHow to use
Paste the text at the start of a chat or into project instructions
Questions about summarizer
How do I install summarizer in Claude Code?
Save the prompt as .claude/commands/summarizer.md.
What does summarizer do?
It condenses an infrastructure optimization search into a factual handoff, retaining candidate checks, metrics, failures, and unresolved hypotheses.
What format does summarizer return?
It returns one JSON object with a summary key. The summary stays within limit_tokens.