Task Activation Protocol

A builder-oriented protocol for turning vague agent promises into traceable, verifiable, checkpoint-driven execution with mandatory post-delivery continuation.

Why builders may care

A common failure mode in agent systems is not intelligence, but execution drift:

• tasks are discussed but never converted into runnable work objects
• no concrete deliverable is defined
• deadlines exist in conversation but not in system state
• there are no checkpoints before the deadline
• stopped tasks can silently resurrect
• delivery does not automatically create the next task

This protocol is designed to plug that gap.

What this package contains

• task_activation_protocol.md — the protocol rules
• task_card_template.json — a structured task card schema
• reactivation_checklist.md — checkpoint-time recovery / shrink-to-MVP checklist
• examples/investment_task_card_example.json — example of a real task card
• examples/post_delivery_review_example.md — example of delivery → review → next task continuation
• UPLOAD_INSTRUCTIONS.md — practical notes for publishing / sharing

Core ideas

Every task must be:

• Traceable — who created it, when, and for what goal
• Verifiable — what counts as done, and how to verify it
• Advanceable — what the checkpoints are, and what happens next
• Valuable — why it matters to the current mainline

Minimum operating rules

1. No task without a task card
2. No task card without:
   • goal
   • value
   • deliverable
   • acceptance criteria
   • verification method
   • checkpoints
   • deadline
3. Every checkpoint must verify a real artifact exists
4. If there is no artifact by a checkpoint, the task must be shrunk into a smaller MVP
5. Every delivery must be followed by review
6. Every review must produce the next task
7. Stopped tasks must be explicitly disabled, not just deprioritized socially

Where this fits in an OpenClaw-style stack

This protocol is a good fit under:

• Skill-based workflows
• Long-running automation loops
• Research / monitoring systems
• Builder-grade agent orchestration
• Multi-step pipelines where “talking about the task” often replaces “finishing the task”

Suggested integration points

Use this protocol when a skill or workflow needs:

• explicit task state
• checkpoint re-activation
• delivery logging
• post-delivery continuation
• protection against stale or stopped tasks reappearing

Example lifecycle

1. Create task card from the JSON template
2. Fill in goal, value, deliverable, acceptance criteria, verification, checkpoints, deadline
3. Start execution
4. At each checkpoint, run the reactivation checklist
5. Deliver an artifact
6. Write a post-delivery review
7. Generate the next task immediately

Good fit for

• OpenClaw builders
• workflow authors
• agent tool / skill authors
• anyone trying to move from “smart conversation” to “reliable delivery”

Not included

This package does not include:

• priority arbitration
• domain-specific planning
• automatic scheduling infrastructure
• execution backends

It is a protocol layer, not a full runtime.

Reuse

Adapt freely for builder projects, private systems, or community workflow packs.