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 skills add secondsky/claude-skills --skill bun-ffigit clone --depth 1 https://github.com/secondsky/claude-skillsWrote 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/skills/secondsky/claude-skills/bun-ffi)<a href="https://agentmods.dev/skills/secondsky/claude-skills/bun-ffi"><img src="https://agentmods.dev/badge/skills/secondsky/claude-skills/bun-ffi/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/secondsky/claude-skills/bun-ffi"><img src="https://agentmods.dev/badge/skills/secondsky/claude-skills/bun-ffi.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- Socket pass
- Snyk pass
- NVIDIA SkillSpector warn
SkillSpector: 1 finding, up to medium
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- medium Memory Poisoning · line 178 Skill attempts to fill the context window with filler content, displacing legitimate instructions and safety constraints. This can degrade agent performance or bypass safety boundaries.Fix: Implement context-window management that detects and rejects padding or stuffing attempts. Prioritize system instructions over user-injected content.
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 5.1 | $0.00056 | $0.02149 |
| Opus 5.5 | $0.00022 | $0.00860 |
| Sonnet 5.5 | $0.00011 | $0.00430 |
| Haiku 4.5 | $0.00006 | $0.00215 |
Grade A, and why
bun-ffi scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 335 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Bun FFI
Bun's FFI allows calling native C/C++ libraries from JavaScript.
Quick Start
import { dlopen, suffix, FFIType } from "bun:ffi";
// Load library
const lib = dlopen(`libc.${suffix}`, {
printf: {
args: [FFIType.cstring],
returns: FFIType.int,
},
});
// Call function
lib.symbols.printf("Hello from C!\n");
Loading Libraries
Platform-Specific Paths
import { dlopen, suffix } from "bun:ffi";
// suffix is: "dylib" (macOS), "so" (Linux), "dll" (Windows)
// System library
const libc = dlopen(`libc.${suffix}`, { ... });
// Custom library
const myLib = dlopen(`./libmylib.${suffix}`, { ... });
// Absolute path
const sqlite = dlopen("/usr/lib/libsqlite3.so", { ... });
Cross-Platform Loading
function getLibPath(name: string): string {
const platform = process.platform;
const paths = {
darwin: `/usr/local/lib/lib${name}.dylib`,
linux: `/usr/lib/lib${name}.so`,
win32: `C:\\Windows\\System32\\${name}.dll`,
};
return paths[platform] || paths.linux;
}
const lib = dlopen(getLibPath("mylib"), { ... });
FFI Types
import { FFIType } from "bun:ffi";
const types = {
// Integers
i8: FFIType.i8, // int8_t
i16: FFIType.i16, // int16_t
i32: FFIType.i32, // int32_t / int
i64: FFIType.i64, // int64_t / long long
// Unsigned integers
u8: FFIType.u8, // uint8_t
u16: FFIType.u16, // uint16_t
u32: FFIType.u32, // uint32_t
u64: FFIType.u64, // uint64_t
// Floats
f32: FFIType.f32, // float
f64: FFIType.f64, // double
// Pointers
ptr: FFIType.ptr, // void*
cstring: FFIType.cstring, // const char*
// Other
bool: FFIType.bool, // bool
void: FFIType.void, // void
};
Function Definitions
import { dlopen, FFIType, ptr, CString } from "bun:ffi";
const lib = dlopen("./libmath.so", {
// Simple function
add: {
args: [FFIType.i32, FFIType.i32],
returns: FFIType.i32,
},
// String function
greet: {
args: [FFIType.cstring],
returns: FFIType.cstring,
},
// Pointer function
allocate: {
args: [FFIType.u64],
returns: FFIType.ptr,
},
// Void function
log_message: {
args: [FFIType.cstring],
returns: FFIType.void,
},
// No args
get_version: {
args: [],
returns: FFIType.cstring,
},
});
// Call functions
const sum = lib.symbols.add(1, 2); // 3
const message = lib.symbols.greet(ptr(Buffer.from("World\0")));
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.
- 1mo ago First seen · 335 lines · 56 tokens per session scan A ba2ce625ea47
bun-ffi is a skill published in the GitHub repository secondsky/claude-skills (236 stars, last pushed 4d ago), licensed MIT. It adds 56 tokens to every session and 2,149 once invoked, about $0.0002 per session on Opus 5.5. 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 2026-09-03.
Other skills, from other repositories
decide
A reusable decision procedure, and a copyable prototype for the guard that encodes it — derive the population, set one threshold by what being wrong costs, name the offender, and break it once to prove it can fail. Use when writing a check, guard, gate or validation of any kind, or when deciding whether something is…
remember
Record why something is the way it is — a decision and its reasoning, a lesson that cost time, or a standing constraint. Use when the reasoning behind a choice would be expensive to reconstruct later.
bootstrap
Set up chamnan in this repository for the first time — build the architecture index, measure how well the code describes itself, fill in missing file comments, and record a baseline. Run once per repo.
milestone
Record a change that reshaped the repository — what moved, why it was worth doing, and which areas it touched. Use after a migration, a rewrite, or a decision that changed how part of the system works.
resume
Write down where this stretch of work stopped, so the next session continues instead of restarting. Use at the end of a working session, or when handing the repository to someone else.
review
Review a change with what this repository knows about it — who depends on the file, which tests cover it, what broke here before, and what the repo has written down. Use before approving a diff, opening a PR, or when asked to review.