intent-to-mission-blueprint

name	intent-to-mission-blueprint
description	Convert human operational intent into a Tandem mission blueprint with phased workstreams, explicit handoffs, reusable outputs, and recurrence-aware setup.
version	1.0.0
tags	["tandem","mission","workflow","orchestration","autonomy"]
compatibility	tandem

Intent To Mission Blueprint

Purpose

Use this skill when a human describes an ongoing operational goal and you need to turn that intent into a Tandem MissionBlueprint.

This skill is specifically for translating human intent into:

• a mission goal
• success criteria
• shared context
• workstreams
• dependencies
• output contracts
• review or approval gates
• recurrence-aware setup

It is not for executing the mission itself.

What Good Output Looks Like

Return one mission blueprint only.

The blueprint should be:

• staged
• scoped
• operationally understandable
• safe for long-running reuse
• concrete enough to compile without hand-waving

The result should feel like something an operator could schedule and trust, not a brainstorming outline.

Core Translation Rule

Translate the user’s intent into the smallest coherent staged mission.

Do not:

• explode one intent into ten weak stages
• collapse everything into one vague workstream
• create review gates everywhere
• treat raw intermediate output as reusable truth

How To Interpret Human Intent

1. Find the real mission outcome

Convert vague requests into one concrete shared outcome.

Example:

• Human intent: "keep the team updated on important operational changes every morning"
• Mission goal: "Produce a reviewed daily operations update with validated changes, recommended actions, and an operator-ready handoff."

2. Find the repeated operating loop

Most long-running missions follow one of a few durable shapes:

• monitor -> analyze -> decide -> handoff
• intake -> plan -> execute -> verify -> review
• collect -> consolidate -> update state -> notify

Choose the smallest loop that matches the intent.

3. Identify the real handoffs

Each downstream stage should consume a clear upstream output.

Use depends_on and input_refs only where the handoff is real.

4. Design for recurrence

If the mission will run daily, weekly, or continuously:

• prefer stable artifacts
• make later stages reuse validated upstream work
• avoid forcing broad rediscovery every run
• include review or approval only where trust actually matters

5. Keep promoted outputs distinct from raw working state

Do not design the mission so every stage treats draft notes as durable truth.

Instead:

• discovery stages produce working artifacts
• synthesis or execution stages produce decision-ready artifacts
• review or approval stages determine what is ready for downstream reuse

Mission Authoring Rules

Goal

The mission goal should describe the end state, not the implementation steps.

Success criteria

Use measurable or inspectable checks.

Good:

• "Final handoff identifies validated changes, required decisions, and clear next actions."

Weak:

• "Mission is helpful and thorough."

Shared context

Include only stable mission-wide constraints:

• audience
• deadlines
• allowed sources
• risk constraints
• compliance or review rules
• quality bar
• cadence assumptions

Workstreams

Each workstream must have:

• one responsibility
• one main artifact or result
• a clear prompt
• bounded scope

Review stages

Add review, test, or approval stages only when they:

• protect downstream trust
• control promotion or delivery
• validate external action readiness
• catch failures that earlier stages are likely to miss

Long-Running Mission Guidance

When the intent implies recurring operation over weeks or months:

• assume project-scoped knowledge reuse
• keep trust floor at promoted by default
• design stages so later runs can build on validated outputs
• avoid requiring every run to rediscover unchanged context
• keep stages inspectable and repairable

Default Staged Patterns

Pattern 1: Monitor -> Analyze -> Decide -> Handoff

Use for:

• operational monitoring
• workflow surveillance
• recurring change detection
• market or environment tracking

Pattern 2: Intake -> Plan -> Execute -> Verify -> Review

Use for:

• execution pipelines
• implementation work
• structured project delivery
• repeated operational processing

Pattern 3: Collect -> Consolidate -> Update State -> Notify

Use for:

• recurring state refresh
• knowledge or system updates
• periodic status synthesis
• publish/update loops

Prompt-Writing Standard For Workstreams

Every workstream prompt should include:

• role
• local mission
• allowed inputs
• output contract
• guardrails

Good prompt skeleton:

Act as the [role]. Use the named upstream inputs and approved workspace context only. Produce the required output in a form the next stage can consume directly. Preserve relevant evidence, avoid redoing upstream work, and record uncertainty or blockers explicitly.

Output Rules

Return valid YAML only.

Include:

• id
• label
• description
• schedule_defaults
• blueprint

The blueprint must contain:

• mission_id
• title
• goal
• success_criteria
• shared_context
• workspace_root
• phases
• milestones
• team
• workstreams
• review_stages

Strong Final Meta-Prompt

Use this when asked to generate a Tandem mission blueprint from human intent:

Convert the following human operational intent into one Tandem mission blueprint.

Requirements:
- Return YAML only.
- Produce one shared mission goal.
- Use the smallest coherent staged mission shape.
- Give each workstream one clear responsibility.
- Use explicit dependencies and input refs only for real handoffs.
- Every workstream must have a concrete prompt and output contract.
- Design for recurring execution when the intent implies daily, weekly, or long-running operation.
- Default to project-scoped promoted knowledge reuse through validated outputs.
- Add review or approval gates only where they materially improve trust, verification, or promotion control.
- Avoid vague stages, vague prompts, and vague outputs.

Human intent:
[insert intent]

Constraints:
[insert constraints]

Preferred workspace root:
[insert workspace root]

Preferred cadence:
[insert cadence]