nathanghart/pathtoship
Production-readiness checks for AI-built apps: scan the repository your builder syncs to, fix what would break, and verify the fix landed.
Changelog
0.4.0 — 2026-09-26
prr_interview: the launch interview as a tool. Thirteen questions no scanner can answer (eight operational attestations + five provider-console figures), recorded per scan and per account, returned beside the score as self_reported and never inside it. In ChatGPT it renders as an inline form (MCP Apps resource). pathtoship-launch-record now records attestations through the tool; pathtoship-readiness points to it before launch; chat instructions gain a section.
0.3.0 — 2026-09-14
check_connection tool: instructions now say to call it when INSTALL_REQUIRED recurs or the connection is unclear.
0.2.1 — 2026-09-14
On INSTALL_REQUIRED, hand over the install link and stop (never scan a different repository instead). Connect guide: claude mcp add runs on the local machine, not a remote session.
0.2.0 — 2026-09-13
Opt-in add-ons: pathtoship-launch-record (launch readiness record — the attestation record) and pathtoship-badge (README/site badge when the verdict is ready). Core pathtoship-readiness unchanged.
0.1.0 — 2026-09-13
Initial pilot kit: chat instructions (Claude.ai / ChatGPT), pathtoship-readiness skill, CLAUDE.md and Cursor rules snippets, connect guide.