drift-probe

Compare proposed artifacts against built state at a given merge SHA.

Updated Jun 4, 2026
One-click install
npx skills add https://github.com/Kohlex/autopilot-plugin --skill drift-probe
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: drift-probe
Source: https://github.com/Kohlex/autopilot-plugin/tree/main/skills/drift-probe
Command: npx skills add https://github.com/Kohlex/autopilot-plugin --skill drift-probe

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Provides a verdict-free data source that answers whether a proposed artifact aligns with what was actually built as of a specific merge SHA, enabling independent downstream decisions at different gates.

Core Features & Use Cases

  • Returns {findings, graph_queried} without applying a verdict, enabling modular gate design and downstream policy.
  • Uses gsd-graphify when available, or falls back to scanning prior phase SUMMARYs to surface drift evidence.
  • Records the evidence path and supports a graph-absent fallback when cross-phase data is not accessible.

Quick Start

Invoke drift-probe with a query and optional graph or prior summaries to obtain evidence about alignment.

Frequently Asked Questions about drift-probe

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

FAQPage Schema
How do I verify whether a proposed artifact aligns with what was actually built after a merge?

Drift alignment verification checks a proposed artifact against the actual build state at a given last-merge SHA, returning structured findings and a graph_queried flag for downstream gates.

What is drift alignment checking and when do I need it in a software engineering workflow?

Drift alignment checking detects misalignment between proposed artifacts and the actual built state at a specific merge SHA, needed when post-merge gates or plan-time drift lenses require evidence of phase consistency.

How do I get drift findings without applying a blocking verdict to my pipeline?

A verdict-free drift probe returns a structured finding set and a graph_queried flag without applying a verdict, allowing downstream gates to independently map the evidence to advisory or blocking outcomes.

Can I check phase alignment if my cross-phase dependency graph is not available?

Yes, the probe falls back to scanning prior phase summaries to surface drift evidence when the graph is absent, recording the evidence path and returning a graph_queried flag for downstream vetting.

Does drift alignment checking require a dependency graph to function?

No, it uses a graph result when available via gsd-graphify, but supports a graph-absent fallback by scanning prior phase summaries to surface drift evidence independently.

What's the best way to use drift findings in post-merge gates and plan-time lenses?

Consume the verdict-free finding set and graph_queried flag directly as a data surface, letting independent downstream gates map the alignment evidence to advisory or blocking outcomes.