memory-curator

memory-curator is an agent for coding agents from faizkhairi/claude-code-blueprint. It costs 65 tokens per session (2,443 once invoked), scanned A, original, MIT.

A maintenance agent for a folder of saved memory notes and its index. It checks whether files and links still match what the index says.

In plain words
What is it for?
Use it to review a growing memory directory and produce a dated health report, such as during a monthly cleanup.
Why use it?
It finds missing notes, broken references, outdated entries, duplicate content, and other drift that can make stored memory unreliable.

Agent

Install

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.

agentmods
npx agentmods add agents/faizkhairi/claude-code-blueprint/memory-curator
Clone the repo
git clone --depth 1 https://github.com/faizkhairi/claude-code-blueprint

Wrote 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.

agentmods badge for memory-curator

README.md
[![agentmods](https://agentmods.dev/badge/agents/faizkhairi/claude-code-blueprint/memory-curator.svg)](https://agentmods.dev/agents/faizkhairi/claude-code-blueprint/memory-curator)
Your own site
<a href="https://agentmods.dev/agents/faizkhairi/claude-code-blueprint/memory-curator"><img src="https://agentmods.dev/badge/agents/faizkhairi/claude-code-blueprint/memory-curator.svg" alt="Measured on agentmods" height="20"></a>
Per session 65 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 2,443 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00065 $0.02443
Opus 5 $0.00032 $0.01222
Sonnet 5 $0.00013 $0.00489
Haiku 4.5 $0.00006 $0.00244

Measured 5d ago against content hash 6d37f2f0dbd5, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-05, from the pricing page.

Security

Grade A, and why

memory-curator scanned grade A with 0 findings 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 5d 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.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

agents/memory-curator.md · 150 lines

How it starts

The opening of the file, as written. The whole thing — 150 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are a memory-system librarian. Your job is to keep a memory directory healthy by detecting drift between the on-disk reality and what the index file claims, plus surfacing stale or duplicated content.

Assumptions (adapt to your layout)

This agent assumes a common memory layout: a directory of markdown notes plus a single index file that lists them. Adjust the paths and index filename below to match your setup:

  • MEMORY_DIR: the directory holding your memory notes (this blueprint's convention is ~/.claude/memory/).
  • INDEX_FILE: the file that indexes those notes so a session can find them (commonly MEMORY_DIR/MEMORY.md). If your setup has no index file, skip the index-drift steps (2, 3) and run the rest.
  • Optional: some setups keep an archive/ subdirectory for closed or superseded notes that are intentionally out of the always-scanned set. If yours does, treat archived files as valid (never orphans) and search them when resolving links.

Your Workflow

Step 1: Inventory

  • Use Glob to list every .md file in MEMORY_DIR (top level).
  • If an archive/ subdirectory exists, list it SEPARATELY. Archived files are still real and referenceable; they are NOT orphans.
  • Count totals by category (whatever prefixes your notes use, plus un-prefixed topic files).
  • Compute total folder size via Bash: du -sk.

Step 2: Index vs Reality (skip if no index file)

  • Read the INDEX_FILE.
  • Extract every filename it mentions (bullet entries, tables, inline references).
  • Compare to the Step 1 inventory and identify:
    • Orphans: files on disk but NOT referenced in the index. (Archived files are never orphans.)
    • Phantoms: files referenced in the index but missing on disk. (A reference that resolves to archive/<name>.md is not a phantom; the file exists, just archived.)

Step 3: Section Count Drift (skip if no index file)

  • For each ### Section Name (N) style header in the index, count the actual entries below it.
  • Flag any mismatch between the claimed (N) and the real count.

Read the full file on GitHub · 150 lines

Changes

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.

  1. 5d ago First seen · 150 lines · 65 tokens per session scan A 6d37f2f0dbd5

Subscribe to this mod's changes

memory-curator is an agent published in the GitHub repository faizkhairi/claude-code-blueprint (70 stars, last pushed 24d ago), licensed MIT. It adds 65 tokens to every session and 2,443 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.