Agent enablement

Permanent link /agent-toolkit/prompt-packs

Prompt packs

Copy these prompt patterns to lead structured teaching sessions and answer common questions in minutes.

Install

Prompt pack v1.0.0

Ethotechnics agent prompt pack

Pin the prompt pack once and reuse it across agent workflows.

Download the pack and keep the version label intact.
Add it to your project at prompts/ethotechnics/ethotechnics-agent-prompt-pack-v1.0.0.md.
Invoke it with use: ethotechnics-agent-prompt-pack-v1.0.0.

Public asset: /agent-toolkit/ethotechnics-agent-prompt-pack-v1.0.0.md

Principal orientation

Summarize the principal's mandate, constraints, and top risks in three bullets.
Name the ethotechnics principle that matters most for the next 30 days and why.
List the top two irreversible decisions in this system and the current exit ramps.

System snapshot

Describe the system in one sentence for a non-technical stakeholder.
Identify the incentive that could undermine ethical outcomes if left unchecked.
Name the data or signal we need to make the system legible.

Release readiness

What would make this release unsafe to ship, even if it meets performance targets?
Which user group bears the largest burden if the system fails?
What rollback path do we have if we discover harm in the first week?

Tradeoff review

State the tradeoff in plain language and who it affects.
Which principle is most threatened by this decision, and how can we offset it?
What evidence would change our minds later?

Values alignment

Name the value conflict in neutral language without assigning blame.
Which principle gives us a shared reference point to resolve the conflict?
What is the smallest decision we can make now that keeps options open?

Escalation check

Does this decision create irreversible harm or regulatory exposure?
Who needs to be informed or sign off to keep accountability intact?
What documentation should we capture for future audits?

Pair these prompts with the glossary to keep terms consistent across teams.

Reference the glossary

Clause violation sweep

Given this system description, enumerate violated STD clauses and cite clause IDs.
List the evidence needed to confirm each violation (logs, receipts, timelines).
Flag any missing inputs that prevent a determination.

Mechanism mapping

Map each violated clause to at least one MEC with justification.
List the artifacts needed to implement each mechanism.
Call out any mechanism coverage gaps or tradeoffs.

Remedy rewrite

Rewrite this policy to satisfy STD-02.C3 with a time-bound reversal path.
Include who owns the remedy and how affected users can contest it.
State how success will be measured (receipt completeness, time-to-remedy).

Expected output format

{
  "system": "short description",
  "violated_clauses": ["STD-01.C2", "STD-02.C3"],
  "evidence_required": ["decision_log", "appeal_endpoint"],
  "findings": [{
    "finding_id": "FND-2026-041",
    "severity": "high",
    "risk": "irreversible harm",
    "recommended_mechanisms": ["MEC-04"]
  }],
  "open_questions": ["missing appeal SLA"]
}

Scoring rubric

Clause IDs are correct and traceable to the standard text.
Evidence requirements are explicit and actionable.
Mechanisms map directly to stated risks or clauses.
Outputs include owners, timelines, or remediation steps.

Common failure modes

Only summarizing values without citing clauses.
Listing mechanisms without evidence or artifact requirements.
Ignoring missing inputs or uncertainty.

Agent Receipt Schema (JSON Schema)

Machine-readable specification for structured decision receipts emitted at runtime.

Defines decision IDs, autonomy levels, and action classes.
Enforces acknowledgment, review, and remedy clocks.

Download JSON Schema

Procurement Clause Pack

Contract-ready legal requirements binding vendor systems to human oversight and contestability.

Binding language for STD-01 / STD-02 enforcement.
Service-credit remedies for unearned closure.

View clause contract

Library search

Prompt packs

Install the prompt pack once