Skip to main content

Prompt History

Prompt history records the metadata needed to understand what was generated before: selected paths, output format, template, estimated tokens, Git metadata, diff mode, output hash, scanner summary, and saved bundle path.

Download for macOSBuy Pro

Why it matters

  • LLM-assisted reviews need a trail of what context was sent and why.
  • Output hashes and selected paths make prior bundles easier to compare without storing raw source contents in history.
  • Prompt history helps teams reconstruct review context after follow-up questions or sponsor fixes.

How it works

  • Record prompt metadata when a bundle is copied or exported.
  • Store a hash of the generated output instead of storing full source contents by default.
  • Redact sensitive values from stored user instructions.
  • Store secret-scan summary metadata, not raw finding values.

Example workflow

  1. Generate a context bundle for an audit question.
  2. Copy or export the bundle after scanner review.
  3. Open prompt history later to inspect selected paths, template, budget, Git state, and output hash.
  4. Use the saved metadata to rebuild or compare a follow-up prompt.

Example history entry

Repo: client-app-review
Selected paths: src/auth/session.ts, src/billing/pricing.ts
Format: markdown
Template: Security review
Estimated tokens: 121600
Output hash: sha256:...
Secret scan: redacted, 2 findings
Bundle saved path: ~/Reviews/client-app/context.md

Limitations

  • History metadata does not replace storing an exported Markdown bundle when a full artifact is needed.
  • If files change after a prompt is generated, restoring the exact source requires the original repository state.
  • Retention and history settings should be reviewed for each team workflow.

Build a repeatable context workflow.

Download the macOS app, review the docs, and contact support for Founder or Pro licensing during launch.

Download for macOSView pricing