Sashiko is an automated code-review system specialized for proposed Linux kernel changes. Kernel maintainers and developers use it to inspect patches from mailing lists or local Git repositories with Linux-specific review instructions and language models. The catalogue add-ons provide its agent workflow.
Borrowing it
Nothing to install: this file belongs to sashiko-dev/sashiko. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/sashiko-dev/sashiko/main/GEMINI.mdgit clone --depth 1 https://github.com/sashiko-dev/sashikoWrote 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/instructions/sashiko-dev/sashiko/gemini-md)<a href="https://agentmods.dev/instructions/sashiko-dev/sashiko/gemini-md"><img src="https://agentmods.dev/badge/instructions/sashiko-dev/sashiko/gemini-md/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/instructions/sashiko-dev/sashiko/gemini-md"><img src="https://agentmods.dev/badge/instructions/sashiko-dev/sashiko/gemini-md.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 | $0.03394 | $0.03394 |
| Opus 5.5 | $0.01358 | $0.01358 |
| Sonnet 5.5 | $0.00679 | $0.00679 |
| Haiku 4.5 | $0.00339 | $0.00339 |
Grade A, and why
sashiko GEMINI.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 — 113 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Role
You're an expert Software Engineer with deep knowledge of Rust, Distributed Systems, Operating Systems and practical experience with infrastructure projects.
Generic guidance
- You MUST commit changes to it after implementing each task or more often if it makes sense. Try to commit as often as possible. Every consistent and self-sufficient change must be committed.
- Sign all commits using the user's git configuration. Every commit MUST include a
Signed-off-byline (e.g., usinggit commit -swhich automatically uses the user'suser.nameanduser.email). NO EXCEPTIONS. Do not use placeholder names or any other default unless explicitly configured in git. - Make sure no lines in the commit message exceed 72 characters. Hard-wrap the commit message body to enforce this length (URLs and quoted log lines that cannot be wrapped are the only exceptions).
- Never use backticks to quote any code, functions and variables names, etc. in the commit message.
- Only standard git trailers belong in a commit message (
Signed-off-by,Fixes,Co-authored-by,Link); never emit tool-generated metadata tags. - When a commit repairs a defect introduced by an identifiable earlier commit on
main, add a Linux-styleFixes:trailer naming that commit:Fixes: <12-char sha> ("<subject>"), generated withgit log -1 --format='Fixes: %h ("%s")' --abbrev=12 <sha>. Identify the introducing commit withgit blameorgit log -S; if it cannot be identified with confidence, omit the trailer rather than guess. If the defect is also tracked as a Sashiko bug, add a second lineFixes: <project>-<uuid>. Trailers form one block at the end of the message, immediately beforeSigned-off-by. - Never start a new task with a non-clean git status.
- Make sure to not commit any logs or temporary files. Before committing, run
make check-pr: it verifiesSigned-off-byon the commit range (RANGE=HEAD~1..HEADby default), runsyamllintonly when YAML files changed, and runsfmt,clippyacross feature profiles, and the tests only when Rust or Cargo files changed. - Every change goes through the two-stage Sashiko-for-Sashiko self-review described under "Development Workflow / 2. Self-Review": the local
sashiko review --project sashiko --agentloop before opening or updating a PR, and acting on everysashiko-botfinding posted on the PR afterwards. - Once the task is done, no local changes should remain. Amend them to the previous commit, if it makes sense, make a standalone commit or get rid of them.
- Each commit should implement one consistent and self-sufficient change. Never create commits like "do X and Y", create 2 commits instead.
- For any non-trivial feature create a design document first, then review it and then implement it step by step.
- If not sure, ask the user, don't proceed without confidence. Also ask for confirmation for any high-level architecture decisions, propose options if applicable.
- Before starting any test or running the main binary, ensure no other
sashikoprocesses are running to avoid port conflicts or database locking issues. - When referring to Sashiko bugs (in PR descriptions, GitHub comments, commit messages, or conversations), always use the public
<project>-<uuid>identifier (e.g.sashiko-<uuid>orlinux-<uuid>). Never refer to bugs by internal SQLite row IDs (Bug #<id>), which are not user-visible and get auto-linked to unrelated GitHub issues or PRs.
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 Changed · +1 lines · +41 tokens per session 7c681cb8c9ad
- 6d ago Changed · +155 tokens per session 440ae132ac4f
- 8d ago Changed · +3 tokens per session 0e483919ba61
- 12d ago Changed · +14 tokens per session 841249062e6b
- 13d ago Changed · -99 lines · -1,282 tokens per session 2b767249da92
- 18d ago Changed · -2 lines · +31 tokens per session c238ca3abf17
- 21d ago Changed · +1 lines · +92 tokens per session 9d14b126f9bc
- 22d ago Changed · +6 lines · +200 tokens per session 5d7598de430b
- 23d ago Changed · +5 lines · +388 tokens per session 409f62f88d6b
- 1mo ago First seen · 201 lines · 3,752 tokens per session scan A 66c332753bb1
sashiko GEMINI.md is an instructions file published in the GitHub repository sashiko-dev/sashiko (1,353 stars, last pushed today), licensed Apache-2.0. It adds 3,394 tokens to every session, about $0.0136 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 2026-08-30.
Other instructions, from other repositories
vscode buildNext.instructions.md
Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).
next.js AGENTS.md
AGENTS.md instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.
vscode oss-third-party-notices.instructions.md
Instructions for microsoft/vscode, covering vs code oss third-party-notices pipeline, architecture, pipeline flow in ci, cache eligibility and rollout and applying the notice (cutover).
langchain AGENTS.md
AGENTS.md instructions for langchain-ai/langchain, covering global development guidelines for the langchain monorepo, corridor security analysis, project architecture and context, monorepo structure and development tools & commands.
vscode copilot-instructions.md
Copilot instructions for microsoft/vscode, covering vs code copilot instructions, project overview, root folders, core architecture (src/ folder) and built-in extensions (extensions/ folder).
deno copilot-instructions.md
Copilot instructions for denoland/deno, covering deno development guide for github copilot, network access, high level overview, key directories and key files to understand first.