ticlawk 0.1.16-dev.18 → 0.1.16-dev.19

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "ticlawk",
-  "version": "0.1.16-dev.18",
+  "version": "0.1.16-dev.19",
   "description": "Local connector that links agent harnesses (Claude Code, Codex, OpenClaw, opencode, Pi) to the Ticlawk mobile app.",
   "type": "module",
   "main": "ticlawk.mjs",

package/src/runtimes/_shared/goal-task-protocol.mjs

@@ -126,11 +126,12 @@ ${buildGoalSetupRules()}
   4. Execute the next task yourself when appropriate, or create/assign a task when coordination is needed.
   5. Check whether the result closes the gap.
   6. Return to evaluating current facts against the goal.
+- During each goal-loop iteration, explicitly write a private loop check before acting: current facts, goal/success criterion, gap, next action, and stop/continue decision. This is for your execution discipline, not a chat message. Do not send the private loop check through ${BRAND_NAME} unless the owner explicitly asks for that analysis.
 - Stop the loop only when there is no meaningful gap, progress is blocked because the owner needs to provide input/resources/permission/confirmation/decision, or progress depends on an external/time-based wait.
 - If there is no gap, say so briefly. If the goal is complete, report completion.
 - If user input, resources, permission, confirmation, or a decision is needed, publish a briefing with one clear bundled request to the owner.
 - If progress depends on an external or time-based future state and no agent or owner can act now, use a reminder or explicit resume condition rather than silently stalling. Do not use reminders to defer an executable next step or an owner decision/request.
-- When a task reaches \`in_review\` or \`done\`, re-run the goal loop before dashboard, MEMORY.md, briefing, or wrap-up updates.
+- When reviewing returned task work, evaluate the evidence against the task and current goal before marking it done or using it in reports. If the task is complete, update task state, then immediately run the private loop check against the updated facts before dashboard, MEMORY.md, briefing, or wrap-up updates.
 - Keep persistent state surfaces distinct: dashboard for goal-level reporting, MEMORY.md for your local continuity, and briefings for active owner notifications.
 - Publish briefings and update dashboards only from DMs you own or groups where you are admin/owner.
 `);
@@ -164,7 +165,7 @@ function buildGroupGoalScopeOverlay() {
 Scope overlay: group with admin or owner role
 - In a group where you are admin or owner, you own both the group goal loop and the task system.
 - Use the group task board for task inventory and assignment.
-- When you delegate substantive work to another agent, make it a group task before or while requesting the work. When results return, update task state and re-run the group goal loop.
+- When you delegate substantive work to another agent, make it a group task before or while requesting the work. When results return, review the evidence, update task state only after task acceptance, then re-run the private group goal loop.
 - Group messages coordinate working agents. Dashboard, MEMORY.md, and briefings are tracking/reporting surfaces with distinct roles.
 - As admin/owner, update the group dashboard when the goal-level report the requester would care about has changed.
 - As admin/owner, publish a briefing only when the owner should be actively notified: milestone reached, important change, blocker, request for owner input/resources/permission/confirmation/decision, or final result.

package/src/runtimes/_shared/standing-prompt.mjs

@@ -31,8 +31,11 @@ private activity text; it is not sent to users or groups.
 - Use the exact target from the current wake message when replying.
 - Complete the work before stopping. Progress updates are allowed, but they
   are not completion.
-- Send messages only when they carry useful content: answer, blocker,
-  decision request, or final result.
+- Use normal assistant output as private work trace. External
+  \`ticlawk message send\` messages should be clean, actionable
+  communication: answer, instruction, blocker, decision request, or final
+  result. Do not send private loop/checklist scratchpad unless the owner
+  explicitly asks for that analysis.
 
 ## Per-Turn Routine