agent-prompt

agent-prompt is a command for Claude Code from jamditis/audiobash. It costs 0 tokens per session (1,153 once invoked), scanned B, original, MIT.

A prompt-improvement command for AudioBash’s voice-to-command mode. It helps shape instructions that turn spoken requests into shell commands while considering the current folder, operating system, shell, recent terminal output, and errors.

In plain words
What is it for?
Use it to improve the agent prompt, add OS-specific command examples, and incorporate custom instructions or vocabulary into speech-to-command handling.
Why use it?
It helps reduce incorrect commands caused by missing terminal context or by using syntax for the wrong operating system or shell.

Command for Claude Code

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 commands/jamditis/audiobash/agent-prompt
Clone the repo
git clone --depth 1 https://github.com/jamditis/audiobash

Made for: Claude Code.

Wrote 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.

agentmods badge for agent-prompt

README.md
[![agentmods](https://agentmods.dev/badge/commands/jamditis/audiobash/agent-prompt.svg)](https://agentmods.dev/commands/jamditis/audiobash/agent-prompt)
Your own site
<a href="https://agentmods.dev/commands/jamditis/audiobash/agent-prompt"><img src="https://agentmods.dev/badge/commands/jamditis/audiobash/agent-prompt.svg" alt="Measured on agentmods" height="20"></a>
Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,153 The whole file, excluding the scripts and references it only reads on demand.
Security scan B 1 finding. 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 $0.00000 $0.01153
Opus 5 $0.00000 $0.00576
Sonnet 5 $0.00000 $0.00231
Haiku 4.5 $0.00000 $0.00115

Measured 4d ago against content hash 317e8bc828ba, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade B, and why

agent-prompt scanned grade B with 1 finding 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 4d 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.

Asks for rootmediumPrivilege escalation

A mod that escalates privileges can change anything on the machine, not only the project.

prompt += `- "permission denied" → Use sudo (Unix) or Run as Admin (Windows)\n`;
.claude/commands/agent-prompt.md · 178 lines

How it starts

The opening of the file, as written. The whole thing — 178 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Agent Prompt Enhancement Generator

You are a prompt engineer improving AudioBash's voice-to-command agent mode. Generate prompt enhancements that improve command accuracy.

Your Expertise

You understand AudioBash's agent mode:

  • Converts natural speech to shell commands
  • Uses terminal context (cwd, shell, OS, recent output, last error)
  • OS-specific command mappings (PowerShell vs Bash)
  • Custom instructions and vocabulary integration

Current Prompt Architecture

// src/services/transcriptionService.ts

private buildAgentPrompt(context?: TerminalContext): string {
  const os = context?.os || 'unknown';
  const shell = context?.shell || 'unknown';

  let prompt = `You are a CLI assistant. Convert speech to executable ${shell} commands.

Current directory: ${context?.cwd || 'unknown'}
Operating system: ${os}
Shell: ${shell}
`;

  if (context?.recentOutput) {
    prompt += `\nRecent terminal output:\n${context.recentOutput.slice(-500)}\n`;
  }

  if (context?.lastError) {
    prompt += `\nRecent error: ${context.lastError}\n`;
  }

  // OS-specific examples
  if (os === 'windows') {
    prompt += `
Use PowerShell commands:
- "list files" → Get-ChildItem
- "go to documents" → Set-Location ~\\Documents
- "show processes" → Get-Process
`;
  } else {
    prompt += `
Use Unix commands:
- "list files" → ls -la
- "go to home" → cd ~
- "show processes" → ps aux
`;
  }

  return prompt;
}

Enhancement Categories

1. Context Extraction

Add new context fields to TerminalContext:

// In electron/main.cjs, enhance getTerminalContext handler

interface TerminalContext {
  cwd: string;
  os: 'windows' | 'linux' | 'mac';
  shell: string;
  recentOutput: string;
  lastCommand?: string;
  lastError?: string;
  // NEW fields:
  gitBranch?: string;      // Current git branch
  projectType?: string;    // node, python, rust, etc.
  runningProcesses?: string[]; // Background jobs
}

2. Prompt Sections

Add new sections to the prompt:

Read the full file on GitHub · 178 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. 4d ago First seen · 178 lines · 0 tokens per session scan B 317e8bc828ba

Subscribe to this mod's changes

agent-prompt is a command published in the GitHub repository jamditis/audiobash (5 stars, last pushed 6d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,153 tokens. A static security scan graded it B with 1 finding (asks for root). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.