git-worktree-runner ai.instructions.md

Instructions for adding support for AI command-line tools to a shell-based project. They explain when to use a shared tool registry and when a tool needs its own adapter file.

In plain words
What is it for?
Use them to register a standard AI tool, create a custom adapter, and update documentation, shell completions, and help text.
Why use it?
They help keep new AI-tool integrations consistent and ensure the tool starts correctly from the chosen project directory.

Instructions file for GitHub Copilot

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/coderabbitai/git-worktree-runner/ai
Clone the repo
git clone --depth 1 https://github.com/coderabbitai/git-worktree-runner

Made for: GitHub Copilot.

Per session 491 This file is loaded in full into every session.
When invoked 491 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 $0.00491 $0.00491
Opus 5 $0.00246 $0.00246
Sonnet 5 $0.00098 $0.00098
Haiku 4.5 $0.00049 $0.00049

Measured yesterday against content hash 046082908fec, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

git-worktree-runner ai.instructions.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 yesterday.

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.

.github/instructions/ai.instructions.md · 59 lines

What it actually says

AI Adapter Instructions

When to Use a File vs Registry

Most AI tools are defined as registry entries in lib/adapters.sh — no adapter file needed. Only create an adapter file in adapters/ai/ for tools that need custom behavior beyond the standard builder (see claude.sh or cursor.sh for examples).

Adding a Standard AI Tool (Registry)

Add a line to _AI_REGISTRY in lib/adapters.sh:

yourname|yourcmd|ToolName not found. Install with: ...|Extra info;More info

Format: name|cmd|err_msg|info_lines (info lines are semicolon-separated)

Adding a Custom AI Tool (File Override)

Create adapters/ai/<name>.sh implementing:

#!/usr/bin/env bash
# ToolName adapter

ai_can_start() {
  command -v tool-cli >/dev/null 2>&1
}

ai_start() {
  local path="$1"
  shift
  if ! ai_can_start; then
    log_error "ToolName not found. Install with: ..."
    return 1
  fi
  (cd "$path" && tool-cli "$@")  # Note: subshell for directory change
}

File-based adapters take precedence over registry entries of the same name.

Also update: README, completions (bash/zsh/fish), help text in lib/commands/help.sh

Contract & Guidelines

  • Must define: ai_can_start (0 = available), ai_start <path> [args...] (runs in subshell (cd ...)).
  • Always quote: "$path" and arguments; never assume current working directory.
  • Use log_error + helpful install hint; never silent fail.
  • Keep side effects confined to worktree directory; do not modify repo root unintentionally.
  • Accept extra args after --: preserve ordering (ai_start receives already-shifted args).
  • Prefer fast startup; heavy initialization belongs in hooks (postCreate), not adapters.
  • When adding adapter: update cmd_help, README tool list, and completions (bash/zsh/fish).
  • Inspect function definition if needed: declare -f ai_start.
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. yesterday First seen · 59 lines · 491 tokens per session scan A 046082908fec

Subscribe to this mod's changes

git-worktree-runner ai.instructions.md is an instructions file published in the GitHub repository coderabbitai/git-worktree-runner (1,754 stars, last pushed 4d ago), licensed Apache-2.0. It adds 491 tokens to every session, about $0.0025 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.