codex-result-handling

Organize Codex helper outputs into structured reviews with verdicts and findings.

Updated Mar 19, 2026
One-click install
npx skills add https://github.com/triodelab/agenci --skill codex-result-handling-triodelab
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: codex-result-handling
Source: https://github.com/triodelab/agenci/tree/main/.claude/codex-plugin-cc-main/plugins/codex/skills/codex-result-handling
Command: npx skills add https://github.com/triodelab/agenci --skill codex-result-handling-triodelab

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill provides a disciplined method to present Codex helper output back to the user by preserving verdicts, summaries, findings, and next steps, enabling auditable reviews and clear communication.

Core Features & Use Cases

  • Preserve the helper's verdict, summary, findings, and next steps structure for review outputs.
  • Present findings first and in order of severity, while keeping evidence boundaries and distinctions such as inferences, uncertainty, or follow-up questions.
  • Preserve all report sections (observed facts, inferences, open questions, touched files, next steps) and, if edits were performed, list the touched files; DO NOT modify code or take actions without explicit user approval.
  • If no findings exist, explicitly state that and keep the residual-risk note brief; for Codex-assisted changes, report updates and defer execution until user confirmation.
  • Special guardrails for codex-rescue: do not auto-implement failed runs; stop and ask the user which issues to fix.

Quick Start

Use a structured Codex output review workflow to preserve verdicts, findings, and evidence boundaries without applying changes automatically.

Frequently Asked Questions about codex-result-handling

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I structure Codex review outputs to preserve findings and evidence boundaries?

Structure Codex review outputs by preserving verdicts, summaries, findings, and next steps, while maintaining evidence boundaries and explicit distinctions like inference or uncertainty. Present findings first in order of severity to enable auditable reviews.

What is the best way to organize Codex helper results into an auditable review?

Organize Codex helper results into an auditable review by preserving all report sections, including observed facts, inferences, open questions, touched files, and suggested actions. Explicitly state when no findings exist and keep the residual-risk note brief.

Can I automatically apply Codex-suggested changes without explicit user approval?

No, you cannot automatically apply Codex-suggested changes. You must report updates and defer execution until explicit user confirmation. Never modify code or take actions without direct user approval, and list any touched files if edits were performed.

Does Codex result handling support failed runs from codex-rescue?

Codex result handling includes special guardrails for codex-rescue outputs. It does not auto-implement fixes for failed runs; instead, it stops and asks the user which specific issues to fix before proceeding.

What sections should a structured Codex output review include?

A structured Codex output review should include observed facts, inferences, open questions, touched files, next steps, and suggested actions. It preserves the helper's verdict and summary while presenting findings in order of severity.

When do I need to direct users to setup instead of processing Codex output?

Direct users to setup instead of processing Codex output when setup or authentication is required. The guardrails enforce routing to the setup process and prevent auto-implementing changes without explicit user approval.