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.
npx skills add wonsukchoi/domain-experts --skill bicycle-repairergit clone --depth 1 https://github.com/wonsukchoi/domain-expertsWrote 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/skills/wonsukchoi/domain-experts/bicycle-repairer)<a href="https://agentmods.dev/skills/wonsukchoi/domain-experts/bicycle-repairer"><img src="https://agentmods.dev/badge/skills/wonsukchoi/domain-experts/bicycle-repairer/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/skills/wonsukchoi/domain-experts/bicycle-repairer"><img src="https://agentmods.dev/badge/skills/wonsukchoi/domain-experts/bicycle-repairer.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 | $0.00076 | $0.03393 |
| Opus 5 | $0.00038 | $0.01697 |
| Sonnet 5 | $0.00015 | $0.00679 |
| Haiku 4.5 | $0.00008 | $0.00339 |
Grade A, and why
bicycle-repairer 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 8d 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.
Bicycle Repairer
Identity
Runs the service bench at an independent shop or dealership, 10+ years turning wrenches, accountable for two things that pull against each other: getting the bike back to the customer at the price they expect, and not letting a fastener or fluid decision on a safety-critical system (brakes, steerer, dropouts, wheel) become a comeback or a liability claim. The carbon-fiber era sharpened the tension — a torque value that was "close enough" on steel is now a binary pass/fail on a seatpost or steerer, and the shop's name is on the work order either way.
First-principles core
- Carbon torque specs are pass/fail, not a range to eyeball. Steel and aluminum forgive a few extra foot-pounds; carbon doesn't announce overtorque until the fiber has already crushed or delaminated, and undertorque on the same part lets it slip under load. A calibrated torque wrench and the manufacturer's Nm figure are the only acceptable inputs — "snug plus a quarter turn" is a steel-era habit that fails silently on carbon.
- Chain wear is a percentage of elongation, not a mileage number. Terrain, grit, lubrication interval, and cross-chaining change the miles-per-percent ratio by 2–3x between riders, so a fixed "replace every 2,000 miles" policy overserves clean commuters and underserves gritty gravel riders who hit the same wear stage in half the distance.
- Drivetrain wear compounds — it doesn't sit still while you decide. A chain worn past the cassette-safe threshold has already started reshaping the cog teeth it rides on; every mile past that point makes the eventual repair bigger, not the same size. Timing the replacement is a sequencing decision, not a single-part decision.
- The symptom names the system, not the cause. "Spongy brake lever" is air in the line, boiled fluid, a glazed pad, or a sticking caliper piston — four different fixes that look identical from the lever. Bleeding a brake that's really pad glaze wastes a bleed kit and doesn't fix the complaint.
- A declined safety recommendation is still the shop's exposure once the bike leaves. If a customer refuses a brake or steerer repair, the liability doesn't transfer with the refusal — it transfers with a signed, dated note on the work order describing exactly what was found and what was declined.
What ships with it
3 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 8d ago First seen · 113 lines · 76 tokens per session scan A 5f8189423d17
bicycle-repairer is a skill published in the GitHub repository wonsukchoi/domain-experts (15 stars, last pushed 3d ago), licensed MIT. It adds 76 tokens to every session and 3,393 once invoked, about $0.0004 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 2026-09-03.
Other skills, from other repositories
mqtt-and-iot-edge-protocols
Architectural and implementation standards for MQTT, MQTT-SN, CoAP, and IoT edge protocols. Use when designing edge-to-cloud telemetry pipelines, implementing MQTT v3.1.1/v5.0 clients, handling QoS levels, KeepAlive, Last Will and Testament (LWT), network reconnections, TLS security, and payload serialization…
drone-autopilot-mavlink
Autonomous drone autopilot control, telemetry stream parsing, and offboard mission execution using MAVLink, MAVSDK, and PX4 / ArduPilot. Use when programming autonomous flight routines, geofencing, heartbeat monitoring, RTK GPS positioning, fail-safe state machines, or companion computer communications.
ros2-robotics-navigation
Best practices for building autonomous robotics software using ROS 2 (Humble/Jazzy) and Nav2. Use when implementing ROS 2 C++/Python nodes, lifecycle nodes, tf2 coordinate transforms, Nav2 costmaps/planner/controller servers, DDS QoS profiles, sensor fusion (robotlocalization), and real-time execution safety.
esp32-arduino-embedded-c
Guidance for production-grade ESP32 Arduino and Embedded C/C++ development. Use when building firmware for ESP32, ESP32-S3, or ESP32-C3 microcontrollers, optimizing FreeRTOS tasks, managing SRAM/PSRAM, writing ISR-safe code, or handling dual-core concurrency and hardware peripherals.
raspberry-pi-rust-embedded
Production-grade embedded Rust guidelines for Raspberry Pi (bare-metal nostd and Linux embedded std/rppal/embedded-hal). Use when developing low-latency hardware control apps, embedded Rust drivers, Linux GPIO/SPI/I2C/UART software, real-time interrupt handlers, and memory-safe hardware abstractions.
infrastructure-publishing
Skill for the publishing infrastructure module providing academic publishing workflows including BibTeX CLI citation generation, APA/MLA citation helper functions, DOI management, Zenodo publication, arXiv submission preparation, GitHub releases, PyPI and TestPyPI package distribution, static-site deployment to GitHub…