Skip to content

TINY_TASKS.md

Latest commit

 

History

History
85 lines (56 loc) · 3.53 KB

TINY_TASKS.md

File metadata and controls

85 lines (56 loc) · 3.53 KB

tinyTasks: Autonomous Task Tracking

tinyTasks is a built-in task management system for tinyMem that lives alongside your code in tinyTasks.md. It acts as an autonomous ledger that ensures AI agents and human operators stay in sync regarding project goals and progress.

🧠 The Philosophy: "Intent lives in the File"

Unlike standard project memory, which can be fuzzy, tinyTasks is authoritative. If it's not in tinyTasks.md, it's not a prioritized task for the agent.

  1. Agent-Led Creation: Agents create and populate tinyTasks.md via task_add when they detect multi-step work. Human verification is optional — agents may ask the user to review the file, but they must not block on it.
  2. Explicit Sync: Every time an agent starts a task or finishes a subtask, it MUST update the file. This ensures the "Project Memory" is always grounded in reality.
  3. Automatic Visualization: The tinymem dashboard reads tinyTasks.md and provides a visual progress report.

🚀 How it Works

1. Initial Creation

If you ask an agent to perform a complex task and tinyTasks.md is missing, the agent calls task_add to create and populate it in one step. The server creates the file automatically.

If you want to review before the agent proceeds, it may ask: "Created tinyTasks.md — verify before continuing?" Respond with yes to review, no / skip to let the agent proceed, or always / never to set a persistent preference.

2. Agent Populates and Proceeds

After creation, the file looks like this:

# Tasks — Implement User Authentication
- [ ] Create database schema for users
- [ ] Implement JWT login endpoint
- [ ] Add unit tests for auth service

The agent begins executing immediately unless you requested verification in step 1.

3. Execution & Tracking

As the agent works, it updates the checkboxes:

# Tasks — Implement User Authentication
- [x] Create database schema for users
- [ ] Implement JWT login endpoint
- [ ] Add unit tests for auth service

When all checkboxes are marked [x], the task is considered complete.

📊 Dashboard Integration

The tinyMem dashboard (tinymem dashboard) provides a real-time view of your task ledger:

  • Completion Rate: Total percentage of finished tasks.
  • Section Breakdown: Progress per major goal.
  • Trend Analysis: How fast tasks are being completed.

🛠 CLI Commands

Command Action
tinymem stats Shows high-level task metrics in the terminal.
tinymem dashboard Opens/Shows the visual status of all tasks.
tinymem health Verifies that the task ledger is correctly synchronized with the memory DB.

📜 Safety Rules for Agents

  • Never infer state: If the file says a task is unchecked, it IS unchecked.
  • Mark completion only when done: Never "pre-check" a task before executing the code.
  • Respect the Ledger: The file is the single source of truth. If the database and the file disagree, the file wins.

FAQ

Q: Can I use multiple task files? A: No, tinyMem specifically looks for tinyTasks.md in the project root to maintain a single, unambiguous source of truth.

Q: What happens if I delete a completed task from the file? A: tinyMem will eventually remove it from its internal analytics as well, but it's better to keep completed tasks for historical context until the goal is fully achieved.

Q: Does it support hierarchical tasks? A: Yes! You can use nested lists to group atomic subtasks under a main task.