disksaver-audit

Audits Grok Bot replica history pressure and verifies archive coverage by comparing entry counts.

Updated Sep 11, 2026
One-click install
npx skills add https://github.com/JYeswak/grok_bot_playground --skill disksaver-audit-jyeswak
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: disksaver-audit
Source: https://github.com/JYeswak/grok_bot_playground/tree/main/plugin/skills/disksaver-audit
Command: npx skills add https://github.com/JYeswak/grok_bot_playground --skill disksaver-audit-jyeswak

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? The Grok Bot desktop keeps only roughly the last 200 transcript entries per Bot and silently drops the oldest with every new turn, so local history can be lost without anyone noticing. This Skill turns the replica set into a pressure verdict: which replicas sit at the client cap, and whether the archive actually holds the entries the cap is about to rotate away. ## Core Features & Use Cases - Pressure read: Lists every replica with its entry count, flags each one at or above the client cap, and records the cap value and pull timestamp. - Archive verification: Confirms archived status by comparing entry counts on both sides (archive vs. replica), never trusting a manifest row, filename, or exit code alone. - Stale-archive detection: Flags archives too old to cover an at-cap replica's missing turns as the top finding, since that is history actively being lost. - Use Case: Before any cleanup that deletes local state, run the audit to get a dated report with a per-replica verdict (safe, at-cap-and-archived, or at-cap-and-exposed) and the single most consequential exposure with its required action. ## Quick Start Ask the assistant to run the disksaver audit against the current Grok Bot replica set and report which replicas are at the cap and whether the archive verifiably holds them.

Frequently Asked Questions about disksaver-audit

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

FAQPage Schema
How do I check if my Grok Bot replicas are losing history?▼

Run a pressure read that lists every replica with its entry count and flags each one at or above the client cap of roughly 200 transcript entries. Record the cap value and pull timestamp, since a count without a cap is not a pressure reading.

How do I verify a Grok Bot replica is actually archived?▼

Verify by comparing entry counts on both sides: archive entries versus replica entries. A manifest row, a filename, or a zero exit code is only a claim, not coverage, so any replica without a count comparison stays on the exposed list.

Why is a stale archive over a capped replica a problem?▼

A stale archive over a capped replica means history is actively being lost, because the gap since the archive's capture holds turns the cap has already rotated away. The audit flags this as the top finding.

Does the disksaver audit delete or modify any replicas?▼

No, the audit is strictly read-only. It never deletes a replica, prunes an archive, rebuilds a Bot, or changes retention behavior; it only reports pressure, archive verification results, and the exposed list.

What happens when a replica cannot be read during the audit?▼

An unread replica is recorded as a gap, not treated as fine. The audit names the replica and states what would settle it, and unreachable surfaces are recorded as unreachable rather than worked around.