Skills Explorer Add Skills

harness-writing

An agent skill by trailofbits, from trailofbits/skills. Tags: code-quality, developer-tools, fuzzing, testing.

What it does

Designs and improves fuzzing harnesses for C/C++ and Rust. Covers mapping raw bytes onto a target API, generating structured inputs, avoiding non-determinism and false crashes, and deciding what to fuzz together. Use when writing a first LLVMFuzzerTestOneInput or fuzz_target! harness, when a campaign finds nothing or reports crashes that will not reproduce, or when the target API needs structured rather than raw input.

Install

With the skills CLI, which installs into Claude Code, Codex, Cursor and other agents:

npx skills add trailofbits/skills --skill harness-writing

Or copy the skill folder into Claude Code's skills directory by hand (~/.claude/skills for every project, or .claude/skills inside one):

git clone --depth 1 https://github.com/trailofbits/skills
cp -r skills/plugins/testing-handbook-skills/skills/harness-writing ~/.claude/skills/harness-writing

Safety box score

Grade A, minimal risk: 7 of 100 risk points, rated 2026-10-17. The grade rates what the skill and its scripts can reach on the machine of whoever installs it, across eight categories from shell execution to secrets access. How the score works.

This is a documentation-only fuzzing skill for C/C++ and Rust. It does not ship runnable automation or secret-handling code, but it does teach the agent to compile and run fuzzers that execute the project under test and generate local build/corpus artifacts.

Category levels, from 0 (none) to 3 (high)
CategoryLevelWhy
Code & shell execution1 (low)The skill gives runnable fuzzing commands like `clang++ -fsanitize=fuzzer,address -g harness.cc -o fuzz_target`, `afl-fuzz ... -- ./fuzz_target`, and `cargo +nightly fuzz run my_target`, but it ships no executables or downloaded scripts.
Network access0 (none)It only links to external documentation and videos; there are no commands that send data off-machine or fetch remote code.
Filesystem reach1 (low)It expects harnesses and corpora to be created inside the repository, e.g. `fuzz/fuzz_targets/`, `seeds/`, and `findings/`, but it does not direct edits outside the project tree.
Secrets & credentials0 (none)There is no mention of tokens, keys, passwords, browser sessions, `.env` files, or other credential stores.
Privilege & persistence0 (none)No sudo, global installs, services, shell startup edits, or OS settings changes are requested.
Instruction hijack surface0 (none)The text contains no prompt-injection patterns such as telling the agent to ignore instructions, hide actions, or follow remote content.
Irreversible actions0 (none)It does not publish, deploy, pay, message, or delete anything; it is limited to local fuzz-harness work.
Transparency0 (none)All shipped files are readable text or SVG; there are no binaries, base64 blobs, or unreviewable installers.

Before installing

Rated 2026-10-17 by an automated review of 3 files from the skill's repository.

Source

Repository
trailofbits/skills (all skills from this repository)
Path
plugins/testing-handbook-skills/skills/harness-writing/SKILL.md
Branch
main
Collection
testing-handbook-skills
Updated
2026-09-19