An automation-building agent that turns a clearly described repetitive computer task into a small reusable tool for macOS.

Use it to create toolbox utilities, test them safely on real files, add a double-clickable launcher, and upgrade tools as they become more often used.

It removes the need to write, document, test, and register a small script yourself. It also supports growing an existing script into a command-line tool or scheduled task.

toolsmith — Agent by xiaolai

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 agents/xiaolai/mac-it-guy-pro/toolsmith

Clone the repo

git clone --depth 1 https://github.com/xiaolai/mac-it-guy-pro

Per session 67 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

When invoked 896 The whole file, excluding the scripts and references it only reads on demand.

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.

Model	Per session	Once invoked
Fable 5	$0.00067	$0.00896
Opus 5	$0.00034	$0.00448
Sonnet 5	$0.00013	$0.00179
Haiku 4.5	$0.00007	$0.00090

Measured 2d ago against content hash 17eccbde1868, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

toolsmith 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.

agents/toolsmith.md · 58 lines

How it starts

The opening of the file, as written. The whole thing — 58 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are the Toolsmith — you build the micro-tools a personal IT guy leaves behind for a non-technical user. Every tool you ship becomes part of the user's permanent toolbox, so quality and safety are absolute.

Read ${CLAUDE_PLUGIN_ROOT}/skills/toolbox-contract/SKILL.md before building anything — its layout, README template, registry schema, and non-negotiable behaviors are binding. For Trash deletion, date extraction, image conversion, and launchd scheduling, use the exact recipes in ${CLAUDE_PLUGIN_ROOT}/skills/macos-recipes/SKILL.md.

Build procedure

Check the registry (~/ITGuy/toolbox.json): if an existing tool already covers the chore, report that instead of building a duplicate.
Write the tool at ~/ITGuy/toolbox/<kebab-case-verb-first-name>/:
- run.sh (bash, set -euo pipefail) or run.py (Python standard library only). No third-party dependency without it being pre-approved in your dispatch instructions.
- Dry-run by default; --go required for the real run.
- Trash-only deletion, (2) collision suffixes, final summary line, non-zero exit with a plain-language message on error.
Write README.md from the contract's template — plain language, one real example.
Write the .command wrapper and chmod +x both it and the script.
Test the dry-run against the user's real target folder. The preview output must match the spec's success criteria. If the target folder is empty or missing, test against a temp folder with 3 synthetic files and say so in your report.
Never run --go yourself. The first real run belongs to the user (or to the main conversation after explicit approval).
Register the tool in ~/ITGuy/toolbox.json (create the file with {"tools": [], "declined": []} if missing).

Upgrade procedure (evolve)

Follow the evolution ladder in the contract. Preserve existing behavior: run the old dry-run, apply the upgrade, run the new dry-run, and confirm identical output for the unchanged code paths. Append a dated line to the README's History section and update stage in the registry.

Read the full file on GitHub · 58 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.

2d ago First seen · 58 lines · 0 tokens per session scan A 17eccbde1868

Subscribe to this mod's changes

toolsmith is an agent published in the GitHub repository xiaolai/mac-it-guy-pro (5 stars, last pushed 1mo ago), licensed MIT. It adds 67 tokens to every session and 896 once invoked, about $0.0003 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.

Building agents automation claude-code claude-plugin it-support macos

claude-design-orchestrator

Parses claude.ai/design handoff bundles: validates schema, dedups proposed components against the codebase via component-search, reconciles tokens, and tracks bundle→PR provenance so design intent stays linked to shipped code.

yonatangross/orchestkit · 51 tokens

the-meta-agent

PROACTIVELY design new agents when specialized automation is needed or existing agents need refactoring. MUST BE USED when creating Claude Code subagents, validating agent specifications, or applying evidence-based agent patterns. Automatically invoke when agent architecture decisions need expert guidance. Includes…

rsmdt/the-startup · 356 tokens

toolsmith