# Launch-Day Buyer Packet

## Buyer

Developers and AI-heavy builders who use Codex, Claude Code, Cursor, OpenCode,
Qwen Code, or other coding agents and need to stop the same agent mistake from
coming back.

## Pain

The painful failure is not "the agent made a mistake." The painful failure is:

- it claimed done without proof
- it touched adjacent business logic
- it added silent fallback behavior
- it skipped or bypassed tests
- it left no prevention artifact for the next run

## Offer

`$19` for one public/redacted failure-to-regression-pack service.

You send:

- intended behavior
- public/redacted agent transcript or diff summary
- public/redacted error or observed breakage
- preferred test runner, if known

You receive:

- failure-mode classification
- deterministic regression checks
- test scaffold suggestions
- guardrail rules
- proof checklist
- next prompt for the next agent or human reviewer

## Why Buy Now

If an AI agent broke one workflow once, the cheap fix is not another confident
agent summary. The cheap fix is a small regression pack that makes the failure
harder to repeat.

## Proof Boundary

No revenue, paid-customer, Product Hunt, or reliability-improvement claim is
valid without evidence in `launch-status.json` and `launch-ledger.md`.

## Public Safety

Public/redacted inputs only. Do not include private repos, private logs,
secrets, tokens, customer data, payment details, private analytics, or NDA
material in public issues.