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 agentmods add rules/savagelysubtle/big-brain-memory-bank/050-reference-verification-systemgit clone --depth 1 https://github.com/savagelysubtle/BIG-BRAIN-Memory-BankWrote 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/rules/savagelysubtle/big-brain-memory-bank/050-reference-verification-system)<a href="https://agentmods.dev/rules/savagelysubtle/big-brain-memory-bank/050-reference-verification-system"><img src="https://agentmods.dev/badge/rules/savagelysubtle/big-brain-memory-bank/050-reference-verification-system.svg" alt="Measured on agentmods" 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 | $0.00000 | $0.02128 |
| Opus 5 | $0.00000 | $0.01064 |
| Sonnet 5 | $0.00000 | $0.00426 |
| Haiku 4.5 | $0.00000 | $0.00213 |
Grade A, and why
050-reference-verification-system 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.
How it starts
The opening of the file, as written. The whole thing — 285 lines — stays where its author put it; the contents beside it link to each section on GitHub.
description: WHEN referencing memory files ENSURE references are verified for accuracy and relevance globs: ["/*.md", "/*.mdc"] alwaysApply: true
TL;DR: The Reference Verification System ensures all references to Memory Bank files are accurate, up-to-date, and relevant to the current context.
1.0.0
**In-line References**
```
According to [ActiveContext:Current Focus], the priority is...
The authentication flow described in [SystemPatterns:Auth Flow] shows...
As noted in [TechContext:Deployment], the staging environment...
```
**Block References**
```
> **Reference**: [FileName:Section]
>
> Relevant content from referenced file...
```
**Table References**
```
| Reference | Key Information | Relevance |
|-----------|----------------|-----------|
| [File1:Section] | Summary of info | How it applies |
| [File2:Section] | Summary of info | How it applies |
```
These consistent formats enable systematic verification.
1. **Existence Check**
- Confirm referenced file exists in Memory Bank
- Verify referenced section exists within file
- If section doesn't exist, find closest match
2. **Recency Check**
- Check last modification date of referenced file
- Verify content hasn't been superseded by newer information
- Note if reference predates significant system changes
3. **Relevance Check**
- Verify information applies to current context
- Confirm reference addresses current task needs
- Assess if more specific/relevant references exist
4. **Accuracy Check**
- Verify quoted/summarized content matches source
- Confirm interpretation aligns with original meaning
- Check for contextual factors that might alter meaning
5. **Extended Context**
- Check for related information in same file
- Identify cross-references to other relevant files
- Note dependencies or prerequisites for understanding
Document verification status using verification indicators.
```
[✓] - Fully verified (exists, current, relevant)
[⚠️] - Partially verified (exists but with caveats)
[⏳] - Pending verification
[❌] - Verification failed (doesn't exist or invalid)
```
For partially verified references, include specific caution notes:
```
[⚠️:outdated] - Information may be outdated
[⚠️:partial] - Reference is partially relevant
[⚠️:ambiguous] - Reference could be interpreted multiple ways
[⚠️:conflict] - Conflicts with other verified information
```
These indicators ensure transparency about reference quality.
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.
- 5d ago First seen · 285 lines · 0 tokens per session scan A a133ce350a86
050-reference-verification-system is a cursor rule published in the GitHub repository savagelysubtle/BIG-BRAIN-Memory-Bank (5 stars, last pushed 1y ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 2,128 tokens. 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 2026-08-31.
Other cursor rules, from other repositories
wrapup
End-of-session wrap-up: summarizes the session and pushes a log to the project's linked NotebookLM notebook. Activate with @wrapup in chat, or "wrap up", "save this session", or "end of session".
memory-bank
You are an expert software engineer with a unique characteristic: your memory resets completely between sessions. This isn't a limitation - it's what drives you to maintain perfect documentation. At the beginning of each dialogue, you rely ENTIRELY on your Memory Bank to understand the project and continue work…
search-routing
Smart Search Routing — always use QMD semantic search when available, never bypass with grep for vault content.
composer-coding-excellence
Coding craft — surgical edits, convention matching, no scope creep, no slop comments, no fabricated APIs.
clarify-first
Infer-and-act by default — ask only on high confusion weight, after inspecting, with a decision-linked question.
cursor
This project uses Gnosys for persistent memory via MCP. Gnosys uses a centralized brain (/.gnosys/gnosys.db) shared across all projects with project, user, and global scopes.