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 instructions/ssx360/matrixscroll/claude-mdgit clone --depth 1 https://github.com/SSX360/matrixscrollWhat 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 | $0.00928 | $0.00928 |
| Opus 5 | $0.00464 | $0.00464 |
| Sonnet 5 | $0.00186 | $0.00186 |
| Haiku 4.5 | $0.00093 | $0.00093 |
Grade A, and why
matrixscroll CLAUDE.md 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.
How it starts
The opening of the file, as written. The whole thing — 89 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md
Always-on rules for writing prose in this repository. The full style guide is a
separate, on-demand Skill (ssx360-prose); this file stays short on purpose,
because it is loaded into every context window.
Run vale --minAlertLevel=error before you start and again before you finish.
The linter gets the last word, never you. See
docs/explanation/writing-workflow.md.
Voice
- Active voice. "The verifier rejects the envelope", not "the envelope is rejected by the verifier".
- Sentence case for every heading. "Exit codes", not "Exit Codes".
- Second person for instructions. "Run the command", not "the user runs".
- Present tense. "The gate blocks the merge", not "will block".
Name real things
Every paragraph must name something a reader could look up: a command, a file, a field, an exit code, a version, a date. A paragraph that only names categories is a paragraph that says nothing. Delete it or replace it with the specific.
Banned words
seamless, seamlessly, cutting-edge, robust, game-changer, unleash, elevate, next-gen, revolutionize, delve, tapestry, leverage, harness, empower, state-of-the-art, best-in-class, world-class, paradigm shift, supercharge, frictionless, turnkey, holistic, synergy, bleeding-edge, military-grade, bulletproof, future-proof, "in today's landscape", "the future of", "deep dive", "rapidly evolving".
Also cut: "it is important to note that", "it should be noted that", "in order to", "utilize", "prior to", "leverage" as a verb.
No em-dashes
Zero em-dashes (—). Zero en-dashes used as a separator ( – ). Use a hyphen, a comma, a period, or parentheses. Use a hyphen for numeric ranges, as in "2024-2025".
One negative parallelism maximum
"X is not Y", "not X, but Y", "It's not X, it's Y", "Notary, not watchtower". One deliberate instance per page, maximum. It is the most recognisable structural AI tell. Vale flags every occurrence as a warning; you decide which one survives.
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.
- 2d ago First seen · 89 lines · 928 tokens per session scan A fa648e485d33
matrixscroll CLAUDE.md is an instructions file published in the GitHub repository SSX360/matrixscroll (5 stars, last pushed 10d ago), licensed Apache-2.0. It adds 928 tokens to every session, about $0.0046 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 instructions, from other repositories
ai-supply-chain-guard AGENTS.md
Instructions for ZarK/ai-supply-chain-guard, covering supply chain guard and public metadata hygiene.
warrant AGENTS.md
Instructions for s0fractal/warrant, covering agent & contributor conduct (governance-critical), hard rules, the human's role and precedent (why this file exists).
agent-runtime-proof AGENTS.md
AGENTS.md instructions for fantasyce/agent-runtime-proof, covering agent runtime proof project rules, product boundary, version 1 scope, evidence rules and authoritative design.
aguara CLAUDE.md
Instructions for garagon/aguara, covering claude.md, project status, build & test commands, single package test and single test function.
uzomuzo-oss project-conventions.instructions.md
Instructions for future-architect/uzomuzo-oss, covering project conventions, test data management, debug with data, not guesswork, general principles and core principles.
uzomuzo-oss ddd-architecture.instructions.md
Instructions for future-architect/uzomuzo-oss, covering ddd layered architecture — strict enforcement, layer dependencies (never violate), layer definitions & responsibilities, responsibility matrix and core domain concepts.