Myco Brain 1.2.5

[1.2.5] — 2026-06-17

A frictionless-install and value-surfacing release. No tool-contract changes —
every brain_* input is unchanged; the new response fields below are additive and
optional. No migrations.

Added

• npx @mycobrain/install — one command to connect any client. A new
  installer (mycobrain-install, fronted by the @mycobrain/install launcher)
  detects your MCP client and writes the right config for Claude Code, Claude
  Desktop, Cursor, Codex, or Windsurf (with --print snippets for Zed, Continue,
  Cline), then runs onboarding. Flags: --client, --all, --print, --scope,
  --no-onboard.
• Onboarding now indexes your own repo (opt-in). On a fresh brain,
  mycobrain-onboard asks before indexing the current project, then proves recall
  on your own code in a fresh context — the fastest felt "it just knew", with no
  export wait. Decline to leave the workspace untouched; --tour runs on
  throwaway sample data instead.
• "Recalled from your memory" attribution. brain_recall_memory and
  brain_context_pack attach an optional attribution credit line that decays as
  the workspace matures (BRAIN_ATTRIBUTION, BRAIN_ATTRIBUTION_DECAY). It rides
  in a structured field, never the result body.
• Pushed stats. brain_context_pack adds a once-per-session session_greeting
  ("Myco has N facts indexed for this workspace"); brain_save_memory adds a
  log-spaced milestone toast (10/50/100/250, then every 250).
• mycobrain-ingest --watch-downloads. Opt-in watcher that auto-imports a
  ChatGPT or Claude export .zip the moment it lands in ~/Downloads.
• Paste-anywhere agent instructions. mycobrain-install prints, and
  docs/agent-instructions.md documents, a portable CLAUDE.md / .cursorrules /
  AGENTS.md contract for when to recall, save, and cite.