Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add KirKruglov/claude-skills-kit --skill memory-auditor-chatgit clone --depth 1 https://github.com/KirKruglov/claude-skills-kitWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/kirkruglov/claude-skills-kit/memory-auditor-chat)<a href="https://agentmods.dev/skills/kirkruglov/claude-skills-kit/memory-auditor-chat"><img src="https://agentmods.dev/badge/skills/kirkruglov/claude-skills-kit/memory-auditor-chat/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/kirkruglov/claude-skills-kit/memory-auditor-chat"><img src="https://agentmods.dev/badge/skills/kirkruglov/claude-skills-kit/memory-auditor-chat.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00101 | $0.02794 |
| Opus 5 | $0.00051 | $0.01397 |
| Sonnet 5 | $0.00020 | $0.00559 |
| Haiku 4.5 | $0.00010 | $0.00279 |
Grade B, and why
memory-auditor-chat scanned grade B with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 10d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Tells the agent never to refusemediumAnti-refusal
Suppressing the ability to decline removes a core safety control; a later harmful request then succeeds.
- If NOT available — switch to read-only audit mode. Analyze only the layers visible in the current context: Project Instructions, Project Summary (if present), Memory Summary (if visible). Inform the user once: "The mem Downgraded: this mod is about security review, or the phrase is quoted, so it is likely naming the pattern rather than instructing it.
How it starts
The opening of the file, as written. The whole thing — 227 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Skill: memory-auditor-chat
Audit Claude's native memory in Claude.ai: detect problematic entries across all available memory layers and propose specific fixes with user confirmation. Works in general chat and projects.
Triggers
Russian: "аудит памяти", "проверь память", "аудит", "почисти память", "разбери память", "что у меня в памяти", "ревью памяти" English: "audit memory", "memory audit", "check memory", "clean up memory", "review memory", "memory health check", "what's in my memory"
Language Detection
Determine the language of the user's first message. Use that language for all subsequent output. If the language is ambiguous, default to Russian.
Input
Explicit user command (trigger). Data for analysis comes from Claude's native memory layers available in the current context.
Output
In-chat report following the fixed Output Template (see below). Layer overview, prioritized issue table, concrete actions: automatic for Memory Edits, manual instructions for other layers.
Reference: Memory Layers and Access Rights
This reference defines the complete and closed list of layers. The skill works ONLY with these layers. Adding new layers, analyzing other sources (file-based auto-memory, CLAUDE.md, User Preferences from Cowork) is prohibited.
| Layer | Description | Skill access |
|---|---|---|
| Memory Edits | Explicit user entries ("remember X"). Scoped: in-project entries are project-only, outside project — global. Limit: 30 entries | Read, remove, replace via memory_user_edits |
| Memory Summary | Auto-synthesized from chat history (~every 24h). Global + per-project | Read only. Provide manual instructions to user |
| Project Summary | Auto-synthesized from project chats. Cannot be edited directly | Read only. Inform user about issues |
| Project Instructions | Manual text in project settings (Edit Project) | Read only. Provide manual instructions to user |
Reference: Context Capabilities
| Context | Available layers | Available actions | Limitations |
|---|---|---|---|
| General chat (Claude.ai) | Memory Edits, Memory Summary | Read all, modify Edits | No project layers |
| Project (Claude.ai) | Memory Edits, Memory Summary, Project Summary, Project Instructions | Read all, modify Edits | Cannot modify Summary/Instructions |
What ships with it
4 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 10d ago First seen · 227 lines · 101 tokens per session scan B 2a2b946ef6e7
memory-auditor-chat is a skill published in the GitHub repository KirKruglov/claude-skills-kit (18 stars, last pushed 1mo ago), licensed MIT. It adds 101 tokens to every session and 2,794 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it B with 1 finding (tells the agent never to refuse). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
memory-management
Two-tier memory system that makes Claude a true workplace collaborator. Decodes shorthand, acronyms, nicknames, and internal language so Claude understands requests like a colleague would. CLAUDE.md for working memory, memory/ directory for the full knowledge base.
ai-agents-architect
Expert in designing and building autonomous AI agents. Masters tool use, memory systems, planning strategies, and multi-agent orchestration.
context-engineering
A project documentation framework that records a project's goals, requirements, roadmap, and current state in separate files. It also keeps those documents within set size limits for an AI coding agent.
compress
Vault alignment engine. Detects and fixes 5 types of structural misalignments: broken backlinks, concept fragmentation, entity miscategorization, duplicated entities, and misnamed entities. Delegates all writes to /bedrock:preserve. Supports interactive mode (user confirmation) and cron mode (autonomous mechanical…
waggle-recall
Recall project-scoped Waggle history about a topic, component, decision, bug, experiment, or implementation attempt.
godot-optimization
Use when optimizing Godot games — profiler, draw calls, physics tuning, memory management, and common bottlenecks.