SlideOps: Verifying Slide Code Synchronization

Solving the Synchronization Problem Between Documentation and Code

Slides or onboarding materials generated by coding agents tend to drift from the actual code immediately after creation. While existing tools excelled at visual completeness, they lacked mechanisms to indicate how outdated documentation had become when code changed. SlideOps is an agent skill that treats generated documents as build artifacts, automatically verifying document freshness when code changes.

Low-Cost Verification Mechanism Based on Hashes

SlideOps records file paths, line ranges, and the SHA-256 hash at that point in time for code citations within slides. The verification process works as follows:

• Detection: The check.py script compares hashes using only the standard library. It requires no model calls, network connections, or token usage, making it virtually free and allowing infinite execution per PR or at session start.
• Repair: It identifies only the mismatched parts (CHANGED) from the verification results and passes them to the agent. Instead of re-reading the entire document, it focuses on modifying only the changed slides, minimizing token costs.

Status Classification and Practicality

Verification results are classified into five states: CURRENT, MOVED, CHANGED, MISSING, and UNVERIFIED. It specifically distinguishes between the MOVED state, where only the code location has shifted, and the CHANGED state, where the logic has changed. This means that for simple location changes, you only need to update the metadata without modifying the prose. It is compatible with major coding agents such as Claude Code, Codex, and Copilot CLI, and complements visual design through role division with existing tools like frontend-slides.