hmem-curate

Audit and clean agent memories with an audit-first workflow.

23|5|Updated Feb 19, 2026
One-click install
npx skills add https://github.com/Bumblebiber/hmem --skill hmem-curate
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: hmem-curate
Source: https://github.com/Bumblebiber/hmem/tree/main/skills/hmem-curate
Command: npx skills add https://github.com/Bumblebiber/hmem --skill hmem-curate

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Memory curations: cleaning, auditing, and consolidating agent memories to ensure governance across runs.

Core Features & Use Cases

  • Process one agent at a time: read, fix, mark audited, summarize, and terminate.
  • Enforce governance by role requirements (e.g., ceo role).
  • Append concise audit summaries and track changes for traceability.

Quick Start

Run a memory curation for the target agent to read, fix, audit, summarize, and terminate.

Frequently Asked Questions about hmem-curate

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

FAQPage Schema
How do I automate agent memory curation and auditing for governance tasks?

Agent memory curation is automated by fetching the audit queue, reading memory to a defined depth, fixing issues, marking audited, and appending an entry to LAST_CURATION.md. This enforces governance and role requirements for a single agent per run.

What is the workflow for cleaning up and consolidating agent memories?

The workflow for agent memory cleanup is a deterministic process: fetch the audit queue, read memory to a set depth, fix identified issues, mark as audited, append a concise summary to LAST_CURATION.md, and terminate. It processes one agent at a time to ensure traceability.

Can I enforce specific role requirements like ceo when auditing agent memory?

Yes, role requirements such as ceo are enforced during the agent memory audit process. The deterministic workflow reads memory to a defined depth and fixes issues while strictly adhering to the governance rules assigned to that specific role.

Does agent memory auditing track changes and provide traceability for governance?

Agent memory auditing provides traceability by appending concise audit summaries to LAST_CURATION.md. This tracks all changes made during the fix phase, ensuring governance compliance and a clear record of what was curated per run.

What are the limitations of running memory curation for multiple agents?

A key limitation of memory curation is that it processes exactly one agent per run. To audit multiple agents, you must execute the deterministic workflow separately for each target, preventing batch processing or concurrent memory consolidation.

Do I need any dependencies to run the agent memory audit workflow?

No dependencies are required to run the agent memory audit workflow. The deterministic process fetches the audit queue, reads memory to a defined depth, fixes issues, and appends to LAST_CURATION.md without external tool dependencies.