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/wez/wiff/agents-mdgit clone --depth 1 https://github.com/wez/wiffWhat 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 | $0.00568 | $0.00568 |
| Opus 5 | $0.00284 | $0.00284 |
| Sonnet 5 | $0.00114 | $0.00114 |
| Haiku 4.5 | $0.00057 | $0.00057 |
Grade A, and why
wiff AGENTS.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 — 53 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Project Overview
wiff is a terminal-centric diff and code-review utility: it captures a diff,
lets you browse and annotate it with syntax highlighting, and persists the
review as a local session that both a human (via the TUI) and an agent (via CLI
and a skill) can read and write concurrently. The binary is wiff.
The design reference is docs/spec.md.
Build Commands
make check # Quick syntax check (cargo check)
make build # Build all binaries
make lint # Clippy, with warnings promoted to errors
make test # Full test suite (cargo nextest run)
make fmt # Format all code (cargo +nightly fmt)
- to rustfmt directly use
cargo +nightly fmt - to run tests directly use
cargo nextest runas it is faster than cargo test. make lintpromotes clippy warnings to errors. Every checkpoint must be clippy-clean: never leave or ignore a warning, even a pre-existing one. If a warning is genuinely not worth fixing, suppress it deliberately with a justified#[allow(...)]rather than letting it ride.
Code Structure
- Rust crates are found in the
cratesdirectory, and are linked into the workspace members list. - Cargo.toml workspace.members must be kept in alphabetical order
- workspace dependencies are used throughout.
- Cargo.toml dependencies are always kept in alphabetical order
- Prefer to use the
anyhow.workspace = trueform when adding a dependency to a crate
Coding Conventions
- Always preserve existing comments when modifying code; both doc comments and inline comments.
- Except in tests, avoid
.unwrap()andpanic!. Prefer to propagate errors using the?operator when in a function that returns aResult. If a panic is unavoidable, use.expect("REASON WHY")instead of a bare.unwrap(). - In tests, always assert on the full rendered output a human would see -- never
examine a substring or a single field in isolation. Do not use
str.contains("something")or other "keyhole" checks; examine the full value. Preferwince::snapshot!orwince::assert_eq!for value comparisons. - If a test has unstable/variable output (eg: temporary file paths), preprocess the string to replace the known temporary path with a constant like TMPDIR before asserting.
- When adding
useimports, place them in a block at the top of the file (or at the top of themod), organized by the formatter.
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 · 53 lines · 568 tokens per session scan A f7b6588f0d45
wiff AGENTS.md is an instructions file published in the GitHub repository wez/wiff (108 stars, last pushed 24d ago), licensed Apache-2.0. It adds 568 tokens to every session, about $0.0028 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
hunk AGENTS.md
Instructions for modem-dev/hunk, covering hunk agent notes, purpose, architecture, shared review seam and architectural rules.
revdiff CLAUDE.md
Instructions for umputun/revdiff, covering revdiff, commands, project structure, architecture principles and config.
GitWand AGENTS.md
Instructions for devlint/GitWand, covering agents.md — gitwand ai agent rules, monorepo structure, security — critical rules, no shell string interpolation in git commands and never bypass saferepopath().
GitWand CLAUDE.md
Instructions for devlint/GitWand, covering gitwand — vue d'ensemble, architecture monorepo, séparation des responsabilités, setup développement and prérequis : node 22.13+, pnpm 11+, rust stable.
blazediff AGENTS.md
Instructions for teimurjan/blazediff, covering blazediff agent rules, hard rules, quick commands and pointers.
critique AGENTS.md
Instructions for remorses/critique, covering opentui, using unreleased opentui versions, bun, react and rules.