feed-the-machine 1.7.17 → 1.7.18

package/ftm-mind/references/orient-protocol.md

@@ -92,6 +92,8 @@ python3 ~/.claude/skills/ftm/bin/brain.py --tasks --task-json
 
 Load active tasks, surface high-priority via TaskCreate. Skip if brain.py absent, tasks loaded recently (15min), or request is purely local.
 
+**NEVER use TaskCreate for execution substeps.** TaskCreate is ONLY for surfacing existing ops tasks from brain.py (the user's real task board). If you're breaking a task into steps, those go in a checkbox plan — not as TaskCreate items that pollute the sidebar. The sidebar shows the user's workload, not your execution state.
+
 ## Playbook Lookup (MANDATORY before external system ops)
 
 **Before any external system operation, check all three knowledge sources:**

package/hooks/ftm-plan-gate.sh

@@ -28,6 +28,16 @@ if [[ "$TOOL_NAME" == "Edit" || "$TOOL_NAME" == "Write" ]]; then
   IS_GATED=true
 fi
 
+# Gate Bash commands that execute scripts or run Python/curl (not simple reads)
+if [[ "$TOOL_NAME" == "Bash" ]]; then
+  COMMAND=$(echo "$INPUT" | jq -r '.tool_input.command // ""' 2>/dev/null)
+  case "$COMMAND" in
+    *python3*|*python*|*node*|*curl*|*"./scripts/"*|*"bash "*|*"sh "*)
+      IS_GATED=true
+      ;;
+  esac
+fi
+
 # Gate mutating MCP calls (create, update, delete, send, add, remove, apply, transition)
 if [[ "$TOOL_NAME" == mcp__* ]]; then
   case "$TOOL_NAME" in

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "feed-the-machine",
-  "version": "1.7.17",
+  "version": "1.7.18",
   "description": "A brain upgrade for Claude Code — 26 skills that teach it how to think before acting, remember across conversations, debug like a war room, run plans on autopilot with agent teams, and get second opinions from GPT & Gemini. Plus 15 hooks that automate the boring stuff.",
   "license": "MIT",
   "author": "kkudumu",