Map the decision
Describe the outcome, constraints, and people affected. A knowledge graph can make dependencies visible when one conclusion relies on several assumptions.
ACOSTA Protocol V3
Fernando Acosta’s roadmap explores how AI systems can combine evidence, explicit rules, and human judgment. The V3 plugin turns that roadmap into a repeatable design review inside Codex.
From roadmap to review
Challenge the proposal
Compare a pattern in the evidence with an explicit rule. Identify where they disagree and what a reviewer would need to resolve the conflict.
Choose a test
Name an observation that would change the recommendation. Record the result so another reviewer can trace the decision.
Try a first review
Suppose a team plans to route support tickets by urgency. A missed urgent ticket and an unnecessary escalation carry different costs. Give V3 the proposed routing rules and ask:
Use ACOSTA V3 to review these rules. Which assumption most affects missed urgent tickets? Give me one test using past tickets and tell me what result would make you change the rule.
Check whether the answer identifies a specific failure mode and a test you can actually run.
Install V3 in Codex
Run this command in your terminal to register the repository.
codex plugin marketplace add fernandoiacosta/ACOSTA-Protocol-v3.0Open the plugin manager, install acosta-v3, then try the review above. The package includes the published roadmap for offline reference.
Open the plugin guideRead the original framework
The historical roadmap includes software-based neuromorphic modeling, neural-symbolic reasoning, human oversight, and peer validation. These are research directions. The published plugin implements the review workflow described above.
Read Fernando Acosta’s annotated roadmap →