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/zealynxsecurity/krait/claude-mdgit clone --depth 1 https://github.com/ZealynxSecurity/kraitWhat 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 | $0.03754 | $0.03754 |
| Opus 5 | $0.01877 | $0.01877 |
| Sonnet 5 | $0.00751 | $0.00751 |
| Haiku 4.5 | $0.00375 | $0.00375 |
Grade A, and why
krait 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 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 — 151 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md
This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
Project shape
Krait is a Solidity security auditor with three delivery surfaces. The first two share patterns and methodology but have separate runtimes; the third is a separate product that happens to live in the same repo.
- Claude Code skill under
.claude/— invoked via/krait,/krait-quick,/krait-fuzz,/krait-review, plus the read-only preflight/krait-init. Pure prompt + methodology files; runs inside the user's Claude session, no API key. Source of truth for the audit methodology. - Standalone CLI under
src/(TypeScript, Node ≥20) — calls the Anthropic SDK directly. Used for batch/shadow audits and CI; needsANTHROPIC_API_KEY. - Checklist plugin under
checklist/— a self-contained Claude Code plugin providing/krait:scan,/krait:assess,/krait:check. Evaluates code against 845 fixed checks across 39 DeFi verticals and emits an importable assessment. Do not apply the audit methodology here: surfaces 1–2 reason adversarially toward exploit traces with a zero-FP bar; the checklist answers "am I audit-ready?" with per-check coverage. Changes to kill gates, heuristics or lenses do NOT propagate to it, and vice versa.
checklist/ was migrated in from the Zealynx-WebSite repo, where it had been developed despite its own plugin.json naming this repo as its home. Its framework JSONs are a synced copy — the website's public/frameworks/ is the source of truth, refreshed weekly from Solodit by a workflow in that repo. Pull updates with checklist/scripts/sync-frameworks.sh <path-to-website>/public/frameworks. The checklist↔website mapping bridge (src/data/krait-checklist-mapping.ts) and its maintainer skill intentionally stayed in the website repo, since they validate against that repo's published data.
Repo root carries .claude-plugin/marketplace.json so /plugin marketplace add ZealynxSecurity/krait resolves — the install path checklist/README.md had always documented but which had no manifest to back it. Install id is krait@krait (plugin krait from marketplace krait); skills are namespaced <plugin>:<skill>, giving /krait:scan, /krait:assess, /krait:check.
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 · 151 lines · 3,754 tokens per session scan A b7df3b29f53d
krait CLAUDE.md is an instructions file published in the GitHub repository ZealynxSecurity/krait (22 stars, last pushed 21d ago), licensed MIT. It adds 3,754 tokens to every session, about $0.0188 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.
Other instructions, from other repositories
suwappubot AGENTS.md
AGENTS.md instructions for 0xSoftBoi/suwappubot, covering agents.md — executable agent policy, ground rules, repo skills (reuse before building), where things live and knowledge base.
quantum-shield CLAUDE.md
Claude Code instructions for kota1026/quantum-shield, covering quantum shield, project structure, universal rules, domain rules and quick start.
clicks-protocol CLAUDE.md
Claude Code instructions for clicks-protocol/clicks-protocol, covering clicks protocol — agent brief, live contracts (base mainnet, owned by safe), wallets, erc-8004 trustless agent and prototypes (repo only, nicht deployed).
ponscli CLAUDE.md
Claude Code instructions for mesutgulecen/ponscli, covering ponscli, the rule that matters most, layout, verifying a change and things that will bite.
PROGRAMMABLE AGENTS.md
AGENTS.md instructions for programmablehq/PROGRAMMABLE, covering programmable workspace rules, canonical repository, parallel work, path ownership and release discipline.
agentic-vault CLAUDE.md
Instructions for agenticvault/agentic-vault, covering agentic-vault, required checks (stop hook enforced), workflow, auto-loop rule ⚠️ and test requirements.