edge-eval

Capture system state, apply invariants, and record evaluation results.

Updated Jan 4, 2026
One-click install
npx skills add https://github.com/DazedtilDawn/operators-edge --skill edge-eval
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: edge-eval
Source: https://github.com/DazedtilDawn/operators-edge/tree/main/codex/skills/edge-eval
Command: npx skills add https://github.com/DazedtilDawn/operators-edge --skill edge-eval

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill provides deterministic evaluation for Operator's Edge by capturing state, verifying invariants, and producing structured reports, reducing manual validation.

Core Features & Use Cases

  • Automates state capture before and after operations to enable reproducible testing.
  • Applies invariant checks and logs evaluation results to a session log for traceability.
  • Generates concise reports summarizing pass/fail status and invariant outcomes to inform maintenance and debugging.

Quick Start

Load active_context.yaml, read evals config, then run edge-eval snapshot to capture the initial state.

Frequently Asked Questions about edge-eval

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

FAQPage Schema
How do I automate deterministic state validation for system invariants?

Automate deterministic state validation by capturing system state, applying invariant checks, and recording results across snapshot, check, run-bank, and report modes to ensure reproducible testing.

What is the best way to capture before and after system state for reproducible testing?

Capture before and after system state for reproducible testing by running a snapshot mode that records the initial state to a before.json file, enabling accurate diff computation later.

How do I generate traceable reports for invariant verification outcomes?

Generate traceable reports for invariant verification by running the report mode, which summarizes pass/fail status and invariant outcomes into structured files within the .proof directory.

Do I need an active_context.yaml file to run edge evaluation snapshots?

Yes, you need an active_context.yaml file containing an evals configuration to run edge evaluation snapshots, as the system reads this config to initiate state capture and write outputs.

Where are evaluation logs and state snapshots stored during an invariant check?

Evaluation logs and state snapshots are stored in the .proof directory, which includes before.json files, eval_run logs, and report summaries to ensure traceable reproducibility.