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 skills/mystenlabs/sui/move-bytecode-comprehensionnpx skills add MystenLabs/sui --skill move-bytecode-comprehensiongit clone --depth 1 https://github.com/MystenLabs/suiWhat 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 | $0.00084 | $0.00958 |
| Opus 5 | $0.00042 | $0.00479 |
| Sonnet 5 | $0.00017 | $0.00192 |
| Haiku 4.5 | $0.00008 | $0.00096 |
Grade A, and why
move-bytecode-comprehension 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 — 64 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Move Bytecode Comprehension
Self-bootstrap: before relying on the routing below, load every reference file — default to
sui prompt skill move-bytecode-comprehension --all, or--list+--file <ref>when budget is tight. This SKILL.md is the mental-model wrapper; reference files cover the binary format and the practice of reading disassembly.
On-chain Sui packages are compiled Move bytecode (.mv, magic 0xA11CEB0B). For
analysis you read it as disassembly (stack-machine, via sui move disassemble). This
skill is the mental model for the binary format and for reading disassembly soundly.
The single most important fact when reading Move bytecode: abilities, visibility, the
entry flag, signatures (incl. generics/phantom), and all on-chain identifiers —
module/struct/field/function names — are preserved. What is lost is human intent
metadata: constant/error names, #[error] messages, local variable names, comments, and
macro structure. Tune reasoning accordingly.
What survives compilation — the survival table
| Source construct | In bytecode? | How it appears in disassembly | Implication |
|---|---|---|---|
| Module address + name | ✓ | module <addr>.<name> |
Identity is exact |
| Struct names | ✓ | struct Name ... |
Type-name reasoning is reliable |
| Field names + types | ✓ | listed on the struct | Field-level reasoning holds |
| Abilities (key/store/copy/drop) | ✓ | has store, key |
Capability / soulbound / hot-potato reasoning is reliable |
| Visibility (private/public/friend) | ✓ | keyword on fn | Call-surface reasoning is reliable |
entry flag |
✓ | entry keyword |
Tx-entry reasoning is reliable |
| Function signatures | ✓ | full (Arg0: T, ...): Ret |
Param/return reasoning holds |
| Generics + phantom type params | ✓ | <T> / phantom |
Type-identity reasoning holds |
| Constant values | ✓ | LdConst[i](T: ...) |
Values readable, names are not |
| Constant / error names | ✗ | LdConst[i] index only |
Use abort codes/positions, not names |
#[error] messages |
✗ | — | Error intent lost; reason from the abort site |
| Local variable names | ✗ | loc0, Arg1, … |
Don't trust names; trust dataflow |
| Comments / doc | ✗ | — | No author intent text |
Macros (assert!, 1u8.do!) |
expanded | inlined opcodes / branches | A macro looks like its expansion |
| Empty struct (e.g. OTW) | ✓ (as 1 field) | dummy_field: bool |
OTW detect via name+drop+synthetic field |
What ships with it
2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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 · 64 lines · 84 tokens per session scan A 9d42803f382a
move-bytecode-comprehension is a skill published in the GitHub repository MystenLabs/sui (7,741 stars, last pushed yesterday), licensed Apache-2.0. It adds 84 tokens to every session and 958 once invoked, about $0.0004 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 skills, from other repositories
squash-bugbot
Triage unresolved bot review comments on a GitHub PR. Use when asked to run /squash-bugbot or to assess, fix, dismiss, reply to, or resolve bot review feedback.
developing-smart-contracts
Solidity smart contract development guidelines for Lineth blockchain. Use when writing, reviewing, or refactoring Solidity contracts, or when the user asks about Solidity best practices, contract structure, or NatSpec docstrings. Covers NatSpec documentation, naming conventions, file layout, and code style.
lineth-quickstart
Operating manual for the Lineth Stack quickstart — the Docker-Compose dev/demo stack at docs/getting-started/lineth-stack in the lineth-monorepo that boots a local Linea/Lineth L2 with Sepolia or local L1 finality. Use whenever you are working inside the lineth-stack quickstart and need to boot or run the stack…
frontend-design
Create distinctive, production-grade Stellar dApp UI. Combines bold aesthetic direction with smart contract integration patterns — wallet connection flows, transaction UX, contract data displays, and agentic kit hook wiring. No generic "AI slop" aesthetics.
stellar-mcp
MCP (Model Context Protocol) tools for Stellar development. Gives Claude direct access to the Stellar blockchain, filesystem, GitHub, and Playwright through configured MCP servers. Use when querying chain data, deploying contracts, managing repos, or running e2e tests from within the agent.
klever-dev
End-to-end Klever blockchain development — smart contracts (Rust/WASM), transaction building (@klever/connect, klever-go-sdk), deployment via ksc + koperator, and on-chain interaction via MCP tools. Use when the task involves KLV, KDA tokens, klv1 addresses, Klever smart contracts, or Klever node/API interaction.