Agents template

Agent Network — Input & Outputs

AgentNet v1.0

Title

[Enter engagement title]

Topic

[e.g., Universal Healthcare / Zoning Reform / Product Strategy]

Requester

[Name or team]

Date

[YYYY-MM-DD]

Priority

[Low / Medium / High / Urgent]

Status

[Draft / In Review / Final]

Input Brief

Objective

One sentence: what “done” looks like.

Audience

Who must act/decide?

Scope & Constraints

Success Criteria

Tags

Use for filtering/search

MethodUse caseFormatAudience

Agents

Architect — Scope & Decomposition

Prompt

Decompose the objective into clear tasks, define deliverables, propose a minimal critical path. Identify needed data and feasible timeline. Output a task table with owner=agent and acceptance criteria.

Deliverables

Task table (owner, input, output, acceptance)
Critical path + milestones
Assumptions & open questions

Output

Connector — Sources & Evidence

Prompt

Find 5–10 high-quality sources across perspectives. Extract the key claims, data points, and dates. Provide formatted citations and note uncertainties/limitations.

Evidence Table

#	Source & Link	Key Claim/Data (with date)	Notes / Limitations
1	[Title] — [Outlet] — [URL]	[Quote or paraphrase + date]	[Bias, sample size, etc.]

Output

Spark — Options & Creativity

Prompt

Generate 3–5 distinct solution options. For each: one-line thesis, why it could work, trade-offs, prerequisites, level of difficulty, time/cost ballpark.

Options Table

Option	Thesis	Pros	Cons/Trade-offs	Pre-reqs
A	—	—	—	—

Output

Mirror — Critique & Stress-Test

Prompt

Stress-test the leading options. Identify hidden assumptions, failure modes, stakeholder objections, and edge cases. Suggest tests or pilots to de-risk.

Risk Matrix

Risk	Likelihood	Impact	Mitigation/Test
—	Low/Med/High	Low/Med/High	—

Output

Guardian — Ethics, Safety, Compliance

Prompt

Evaluate options against values (good-faith, transparency, proportionality, dignity), legal/compliance constraints, and reputational risk. Recommend guardrails and redlines.

Guardrails Checklist

Transparency (sources, incentives, uncertainties)
Privacy / data handling
Fairness / non-harm / accessibility
Governance (approvals, audits, logs)

Output

Integrator — Synthesis & Decision

Prompt

Synthesize agent outputs into a single recommendation. Include rationale, selection criteria, success metrics, simple timeline, owners, and next 3 actions.

Decision Snapshot

Chosen Option	Why (criteria)	Metrics	Next 3 Actions
—	—	—	1) … 2) … 3) …

Output

Handoff Package

Deliverable	Format	Owner	Status
Executive Summary	HTML/PDF	Integrator	[Draft]
Evidence Appendix	Links/CSV	Connector	[Draft]
Risk & Guardrails	Table	Mirror/Guardian	[Draft]

Note: Use consistent file names (YYYYMMDD_topic_v1.pdf) for versioning.

Export (JSON)

{
  "title": "[copy from meta_title]",
  "topic": "[copy from meta_topic]",
  "requester": "[copy from meta_requester]",
  "date": "[copy from meta_date]",
  "priority": "[copy from meta_priority]",
  "status": "[copy from meta_status]",
  "objective": "[copy from in_objective]",
  "audience": "[copy from in_audience]",
  "constraints": "[copy from in_constraints]",
  "success_criteria": "[copy from in_success]",
  "tags": "[copy from in_tags]",
  "agents": {
    "architect": {"prompt": "[architect_prompt]", "output": "[architect_output]"},
    "connector": {"prompt": "[connector_prompt]", "output": "[connector_output]"},
    "spark": {"prompt": "[spark_prompt]", "output": "[spark_output]"},
    "mirror": {"prompt": "[mirror_prompt]", "output": "[mirror_output]"},
    "guardian": {"prompt": "[guardian_prompt]", "output": "[guardian_output]"},
    "integrator": {"prompt": "[integrator_prompt]", "output": "[integrator_output]"}
  }
}

Tip: If you use ACF/Meta Box, map these fields to custom meta for programmatic exports.