Skip to content

thumbgate@1.27.6

Choose a tag to compare

@github-actions github-actions released this 05 Jun 19:12
b1e0d7c

thumbgate@1.27.6

Release Links

npm Email Companion

npm controls the native "Successfully published" email template, so the email itself stays short. Treat this generated artifact as the full release-note companion for that email: it carries the Changeset summaries, CHANGELOG entry, publish workflow, npm tarball, and shasum when available.

Full Changeset Release Notes

No changed .changeset/*.md entries were detected for this release range.

CHANGELOG.md Entry

1.27.6

Patch Changes

  • #2505 4c24b82 Thanks @IgorGanapolsky! - Fix /checkout/pro zombie sessions: require customer_email on POSTs. The earlier 401 fix unblocked POST traffic, but 3 of 4 fresh POST-flow sessions then reached Stripe with customer_email=null, creating un-recoverable zombie sessions (no email = no recovery campaign possible). Now POSTs without a customer_email query param fall through to the email-capture interstitial instead of auto-confirming. GET-with-confirm=1 behavior is unchanged.

  • #2510 1d51f56 Thanks @IgorGanapolsky! - dashboard: make "Chat with your data" intent-aware (lists, time windows, real listings)

    The local-first chat (#2501) routed every question to one of 5 canned per-topic
    paragraphs — so "how many gates are activated today?" and "what mistakes were
    blocked today?" both returned the same Active gates: N. Blocked: M. line.
    Two underlying bugs:

    1. Classifier order — /block/ hijacked "what mistakes were blocked today" into
      the gates topic. Feedback-specific words ("mistake", "lesson", "feedback",
      "thumbs", "negative", "positive", "wins", "what went wrong") now run FIRST.
    2. Section answers ignored intent — "what" got the same line as "how many", and
      "today" was never filtered. Now the section builder parses intent
      (wantsList vs count, windowMs: today/yesterday/this-week/this-month) and
      reads the feedback log directly to list real mistakes/wins for the requested
      window, instead of a canned all-time count.

    Examples that now answer distinctly (all still local, no cloud):

    • "how many mistakes today?" → Feedback today: 3 (1 positive, 2 negative).
    • "what mistakes today?" → enumerated list of today's negative contexts.
    • "show me wins this week" → enumerated list of positive entries from 7d.
    • "what gates do we have?" → enumerated list of active gates with severity.
  • #2505 4c24b82 Thanks @IgorGanapolsky! - Fix dashboard "Chat with your data" for keyless local installs: the exact metric question "how many mistakes were prevented today?" is now covered by the local-data regression test with every cloud/model env var removed, and the dashboard copy no longer implies Gemini/Perplexity keys are required for local answers.

  • #2505 4c24b82 Thanks @IgorGanapolsky! - Improve local-first dashboard chatbot to answer specific queries like "what mistakes did I make?" or "what mistakes were blocked today?" by scanning the local feedback log and listing the 5 most recent mistakes/actions instead of returning a generic summary.

  • #2507 3e6ccb4 Thanks @IgorGanapolsky! - fix(mcp): stdio server self-exits on stdin EOF/close

    stdio MCP server (adapters/mcp/server-stdio.js) now listens to stdin 'end' and 'close' events to exit the process with code 0 when the client disconnects. This prevents orphaned serve processes from accumulating on the host system.

  • #2512 deb3355 Thanks @IgorGanapolsky! - Fail closed when live Stripe webhook processing is enabled without STRIPE_WEBHOOK_SECRET; unsigned webhook parsing is now test-only opt-in.

Verification Standard

  • Publish only runs from main after version sync, tests, and runtime proof pass.
  • The npm package is smoke-tested after publish by installing thumbgate@VERSION in a clean runtime.
  • GitHub Release notes are generated from Changesets, not only GitHub auto-generated PR titles.