git-authoring GEMINI.md

Repository instructions for git-authoring, a project that defines how to prepare commits, pull requests, reviews, merges, and release notes. A pull request is a proposed change submitted for review before merging.

In plain words
What is it for?
They help inspect repository history, write commit messages and pull-request text, prepare release notes, and handle branch or merge work.
Why use it?
They keep Git contributions consistent and ensure repository actions match the user’s requested level of control.

Instructions file for Gemini CLI

Install

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.

agentmods
npx agentmods add instructions/n-shadloo/git-authoring/gemini-md
Clone the repo
git clone --depth 1 https://github.com/n-shadloo/git-authoring

Made for: Gemini CLI.

Per session 1,448 This file is loaded in full into every session.
When invoked 1,448 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.01448 $0.01448
Opus 5 $0.00724 $0.00724
Sonnet 5 $0.00290 $0.00290
Haiku 4.5 $0.00145 $0.00145

Measured 2d ago against content hash 69e118640b60, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

git-authoring 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.

GEMINI.md · 87 lines

How it starts

The opening of the file, as written. The whole thing — 87 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Git authoring conventions

This repository's commit-message and pull-request rules live in AGENTS.md at the repository root. Follow it in full and choose the mode from the user's request:

  1. For already-staged changes, present the exact quoted-heredoc commit command.
  2. When asked to choose files, present specific staging commands plus the heredoc commit command.
  3. When asked for PR content, write the title and structured Markdown description.
  4. Only when unmistakably told to do it all, stage, commit, and push autonomously — and, on a further explicit request, tag and publish a release.
  5. When asked for release notes, establish the range since the last release, read what actually landed, and write the note as a Markdown block.
  6. When asked to review an incoming pull request, read it through gh, work the decision through with the user, and write the review comment or the merge-commit message as a Markdown block for them to paste. Never approve, request changes, comment, or merge.
  7. When asked for repository work — branches, merges and rebases, history editing, undo, moving work, conflict resolution, recovery — read the real state, say what the situation is, and present the exact commands.

Modes 1–3 and 5–7 are always read-only, even after a follow-up confirmation. Mode 4 is the sole execution exception, and mode 7 executes only under that same explicit request; it is an exception to the read-only rule and to nothing else. Never infer mode 4 from "commit this," "go ahead," or a request for commands. Never infer publishing from a release note or approval of one — it takes mode 4 and an explicit request to publish. Never use git add -A, force-push, or overwrite an existing tag or release. Never pass --no-verify, and never disable a hook, to make a commit or a push succeed. A hook that fails is a finding, and it stops the work.

Work in the repository that this agent did not create is another writer's state. Inspect it, never sweep it into a commit, and never rewrite it. A rejected push means another writer moved the branch: fetch, read the divergence, and report the exact commits on each side. Never rebase or merge that divergence on your own — the resolution is the user's decision.

Every stop hands the work back with five fields. The goal, and the exact blocked step. What you attempted, and the git output, verbatim. The causes you eliminated, and how. The one decision needed from the user. The state that remains, and whether it is safe to leave. Mode 4 claims completion only from git log -1 --format=full and git status. Run both after the push. A claim without that output is not a completion.

Mode 4 does not extend to mode 6. Deciding to approve, reject, or merge someone else's pull request always produces text the user pastes, never an action the agent takes — "approve it" or "yes, merge" tells you what the block should say, not to run it. Never run gh pr review, gh pr comment, gh pr merge, or gh pr close.

When reviewing a pull request, separate what actually blocks the merge — breaks the build or an existing test, loses or corrupts data, opens a security hole, breaks a documented contract without notating it, or does not do what the PR claims — from what is only a suggestion, and calibrate the bar to the project: a small repo with no stated convention and no CI is not a large one. Never manufacture a blocker, and when nothing blocks, say so in one plain line before anything else. On a fork pull request, empty gh pr checks output means the workflows have not run, not that they passed.

For a release note, fetch tags first, base the range on the last version tag reachable from HEAD, and report any disagreement with the project's version field rather than silently picking one. If the repository has no versioning at all, say so and stop rather than creating its first tag. Match the style of prior releases over any template.

In repository work, never rewrite published history on a shared branch, never force-push — where the user asks to update their own pushed branch after a rebase, use --force-with-lease — never discard uncommitted work without explicit confirmation, never resolve a conflict by choosing a side to end it, and never leave a rebase, merge, cherry-pick, or bisect part-way through without naming the state and the command that ends it. Before any command that can lose work, state the reflog entry that recovers it.

Read the full file on GitHub · 87 lines

Changes

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.

  1. 2d ago First seen · 87 lines · 1,448 tokens per session scan A 69e118640b60

Subscribe to this mod's changes

git-authoring GEMINI.md is an instructions file published in the GitHub repository n-shadloo/git-authoring (4 stars, last pushed 6d ago), licensed MIT. It adds 1,448 tokens to every session, about $0.0072 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 2026-08-31.

Related

Other instructions, from other repositories

claude-codex-settings CLAUDE.md

Claude Code instructions for fcakyon/claude-codex-settings, covering claude-settings, repo structure, cross-tool plugin format reference, plugin manifests and root marketplace files.

fcakyon/claude-codex-settings · 4,944 tokens

claude-codex-settings AGENTS.md

AGENTS.md instructions for fcakyon/claude-codex-settings, a project described as: Battle-tested Claude Code, OpenAI Codex, Cursor configs, plugins, hooks and agents with Kimi, MiniMax and GLM API support.

fcakyon/claude-codex-settings · 4 tokens

claude-codex-settings GEMINI.md

Gemini CLI instructions for fcakyon/claude-codex-settings, a project described as: Battle-tested Claude Code, OpenAI Codex, Cursor configs, plugins, hooks and agents with Kimi, MiniMax and GLM API support.

fcakyon/claude-codex-settings · 4 tokens

awesome-copilot-id AGENTS.md

Instructions for GulajavaMinistudio/awesome-copilot-id, covering communication, explanation and documentation, markdown formatting, user communication style and workflow & methodology.

GulajavaMinistudio/awesome-copilot-id · 7,686 tokens

hiveshare CLAUDE.md

Instructions for KB-perByte/hiveshare, covering hiveshare — claude.md, repo layout, build & run, key env vars (server) and naming: hive vs memory.

KB-perByte/hiveshare · 1,301 tokens

llm-safe-haven CLAUDE.md

Instructions for pleasedodisturb/llm-safe-haven, covering llm safe haven, what this is, project structure, tdd — non-negotiable (adopted 2026-08-17) and the contract.

pleasedodisturb/llm-safe-haven · 1,322 tokens