dual-memory

A project rule requiring two kinds of memory before substantial code changes: semantic search for relevant knowledge and structural analysis for code relationships.

In plain words
What is it for?
Preparing features, refactors, debugging, architecture decisions, dependency changes, and non-trivial reviews by querying Qdrant memory and the project's structural graph.
Why use it?
It helps agents find prior decisions, known problems, and usage patterns before editing, reducing changes based on incomplete project understanding.

Cursor rule for Cursor

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 rules/dzmitrys-dev/supamem/dual-memory
Clone the repo
git clone --depth 1 https://github.com/dzmitrys-dev/supamem

Made for: Cursor.

Per session 208 This file is loaded in full into every session.
When invoked 208 The same file — it is already loaded in full.
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 $0.00208 $0.00208
Opus 5 $0.00104 $0.00104
Sonnet 5 $0.00042 $0.00042
Haiku 4.5 $0.00021 $0.00021

Measured 2d ago against content hash 4737e6d4acd8, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

dual-memory 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 2d 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.

.cursor/rules/dual-memory.mdc · 34 lines

What it actually says

Dual Memory — supamem

Use both memory layers before any non-trivial code change.

When to query

  • New feature, refactor, debugging, or architecture decision
  • Adding/modifying dependencies
  • Code review of a non-trivial change

Semantic (Qdrant tuned_hybrid)

qdrant-find "<topic> decisions"
qdrant-find "<topic> known issues"
qdrant-find "<topic> patterns"

Refresh

If memory feels stale: supamem index --target tuned --force.

Anti-patterns

  • ❌ Querying after coding — check memory before implementing.
  • ❌ Skipping semantic for "what calls X" — use structural for that.
  • ❌ Trusting hits without checking source attribution — always inspect the payload's source field.
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. 2d ago First seen · 34 lines · 208 tokens per session scan A 4737e6d4acd8

Subscribe to this mod's changes

dual-memory is a cursor rule published in the GitHub repository dzmitrys-dev/supamem (12 stars, last pushed 12d ago), licensed MIT. It adds 208 tokens to every session, about $0.0010 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.