RPCS-1

examples

High-intent calls to try first.

These are the cases where RPCS-1 is most obvious: a support copilot under pressure, a coding agent in a changing repo, or a research agent handling conflicting evidence. The first useful answer should lead with TI, SG, FT, UE, AR, then the failure-risk score and next test.

Support copilot under pressure

Use this when refunds, billing disputes, policy exceptions, and queue pressure make the agent behave inconsistently.

Run in the tuner

example request

Use recommend_agent_configuration to diagnose my support copilot.

Task: refund and billing dispute triage
Environment: dynamic, somewhat_predictable, high stakes
Context relevance: medium
Commitment style: cautious
Target platform: anthropic

Coding agent in a changing repository

Use this when a coding agent retries too aggressively, changes direction, or commits before it has enough context.

Run in the tuner

example request

Use recommend_agent_configuration to diagnose my coding agent.

Task: inspect a changing repository, edit files, run tests, and open a pull request
Environment: moderate, somewhat_predictable, medium stakes
Context relevance: long
Commitment style: balanced
Target platform: openai

Research agent with conflicting sources

Use this when a research agent overreacts to new sources, loses earlier evidence, or sounds too confident.

Run in the tuner

example request

Use recommend_agent_configuration to diagnose my research agent.

Task: synthesize conflicting technical sources into a cautious recommendation
Environment: stable, highly_predictable, medium stakes
Context relevance: long
Commitment style: cautious
Target platform: generic

What the output should lead with

  • • TI, SG, FT, UE, AR
  • • Failure-risk score
  • • Predicted regime
  • • Runtime posture
  • • Best next test