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/ssheleg/super-ux/visiongit clone --depth 1 https://github.com/ssheleg/super-uxWhat 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 | $0.00044 | $0.00748 |
| Opus 5 | $0.00022 | $0.00374 |
| Sonnet 5 | $0.00009 | $0.00150 |
| Haiku 4.5 | $0.00004 | $0.00075 |
Grade A, and why
vision 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 yesterday.
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 — 68 lines — stays where its author put it; the contents beside it link to each section on GitHub.
vision — what the product is, and what it refuses to become
The layer above the chain. foundation.md answers who uses the product
and why; vision.md answers what the product is. Do not confuse it
with scenarios.md, the other layer whose name starts with "what":
scenarios say what the product does, step by step. A feature can satisfy
every scenario and still violate the anti-vision — that case is the whole
reason this layer exists.
Before writing a word
Read the project, not its README: source entry points, the schema, the
routing table, the live UI, package.json, and docs/ux/foundation.md if
it exists. A vision written from the README alone describes the README.
The gap between what a project claims and what its code does is usually
where the real vision is hiding, and naming that gap is often the most
valuable thing this pass does.
The document — nine sections, these headings, in this order
## 1. Essence one sentence; survives a rewrite on another stack
## 2. Core idea an observation about the world, not a problem statement
## 3. What the system does continuous behaviour, verbs not nouns
## 4. The user's role what the user becomes: directs, judges, decides
## 5. Principles "we do X, not Y" — Y is something reasonable people pick
## 6. Anti-vision what it refuses to become, named product by product
## 7. Horizon a direction for two to three years, never a roadmap
## 8. The one sentence if someone remembers exactly one thing
## 9. The alignment test three to five questions a feature must pass
**Status:** draft | approved and **Last reviewed:** YYYY-MM-DD at the
top. A section with no rejected alternative is a slogan; a vision with no
anti-vision settles no argument.
The half that is actually the point
Writing the document is half the job. The other half is the alignment rule
in this project's own instruction file (.cursor/rules/, AGENTS.md or
CLAUDE.md — whichever this project reads), so that later features get
checked against the vision instead of the vision being written once and
never read. A vision nothing reads is a document, not a constraint.
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.
- yesterday First seen · 68 lines · 44 tokens per session scan A 135249f03e42
vision is a cursor rule published in the GitHub repository ssheleg/super-ux (2 stars, last pushed 2d ago), licensed MIT. It adds 44 tokens to every session and 748 once invoked, about $0.0002 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 2026-08-31.
Other cursor rules, from other repositories
tell-proof-verify
Auto-run Tell Proof verification after UI-changing edits.
tell-mission
Tell mission and golden rules — always apply.
tell-plumbing-reference
Peer plumbing reference memory for platform/MCP/auth/design-flow plans — never name the peer in commits.
tell-domain-research
Auto-trigger website domain research before any new site build.
tell-redesign
Redesign and reconciliation rules.
tell-schema-contracts
Zod schema contracts — packages/schema.