Skip to content

v3.0.75 — foolproof install & auth for AI agents

Choose a tag to compare

View all tags
@chandshy chandshy released this 08 Jun 17:09
· 2 commits to main since this release
v3.0.75
e9e78fa
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Closes the "an AI can't recognize, install, or connect this MCP server" gap.

Added

  • MCP server instructions on the initialize response (both stdio and HTTP transports) — identity + connect lifecycle (call setup_status first; what unconfigured/bridge-unreachable/pending-approval/ready mean and the exact next action). Previously empty.
  • setup_status tool — always-available, ungated, read-only install/connect diagnostic (dispatched pre-gate like the escalation meta-tools). Returns a state machine + the single next step; visible at every tier. Redacts the mailbox address and config path for callers whose grant isn't yet active.
  • mailpouch setup — non-interactive credential setup (--username, --password-stdin/--password-file/--password). Writes through the account registry so the secret lands on the authoritative per-account keychain key.
  • mailpouch doctor — prints the install/connect diagnosis from a shell (--json); exits non-zero until ready.

Changed

  • Single canonical MCP client config across the docs: { "command": "npx", "args": ["-y", "mailpouch"] }.
  • llms.txt leads with an AI quickstart; README/README_FIRST_AI gain a "zero to running" block; README_FIRST_AI.md + llms.txt now ship in the npm tarball.
  • Tool count reconciled to 76 (73 categorized + 3 always-available meta); 127.0.0.1 in all connection tables.

Full changelog: https://github.com/chandshy/mailpouch/blob/main/CHANGELOG.md

Assets 2
Loading