Borrowing it
Nothing to install: this file belongs to GabrielOnDelphi/Claude-Tools-for-Delphi. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/GabrielOnDelphi/Claude-Tools-for-Delphi/main/.claude/skills/light-bug/SKILL.mdgit clone --depth 1 https://github.com/GabrielOnDelphi/Claude-Tools-for-DelphiWrote 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/gabrielondelphi/claude-tools-for-delphi/light-bug)<a href="https://agentmods.dev/skills/gabrielondelphi/claude-tools-for-delphi/light-bug"><img src="https://agentmods.dev/badge/skills/gabrielondelphi/claude-tools-for-delphi/light-bug.svg" alt="Measured on agentmods" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00203 | $0.03108 |
| Opus 5 | $0.00102 | $0.01554 |
| Sonnet 5 | $0.00041 | $0.00622 |
| Haiku 4.5 | $0.00020 | $0.00311 |
Grade A, and why
light-bug 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 today.
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 — 133 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/light-bug — Delphi bug diagnosis and fix (general entry point)
One skill for "there's a bug", regardless of how it arrived — a Thunderbird crash report, a pasted stack trace, or three sentences describing wrong behavior. Intake -> reproduce -> localize -> derailment check -> fix -> verify, every time. The derailment check is what stops a plausible-looking but wrong root cause from turning into a wasted afternoon.
You (the main thread) run this directly. Unlike the crash-report pipeline there is no workhorse agent: the derailment check needs to happen where the conclusions were actually drawn. A subagent's report only counts as INFERRED until re-checked, so relaying the reasoning secondhand would defeat the point of checking it at all.
Step 0 — Route
Work through these in order — the first one that matches decides the path. Do not skip ahead to "general path" just because a message isn't obviously a crash report; case 3 below exists precisely to catch the messages that are too thin to act on either way.
- Unambiguous crash-report evidence — a
.madfile path, or wording like "crash report" / ".mad" / "bug report email" / "madExcept". This is the crash-report pipeline: readc:\Users\<you>\.claude\skills\light-bug-MadShi\SKILL.mdand follow it exactly (it resolves the product, lists/extracts the.mad, and launches its own agent). That skill is self-contained end to end — do not continue into Phase 1 below. - An Android-only symptom (works on Windows, fails only on the phone; "disappears on the phone"; a logcat paste) — hand off to the
light-bug-Androidskill (Skill tool) instead of continuing here. - Too thin to act on — neither 1 nor 2 matched, AND you cannot pin down BOTH (a) which product/project this is about and (b) a concrete symptom (the message is just "there's a bug" / "it crashes", names no registered product, and no project is obvious from the working directory). Do not guess on either axis — ask with
AskUserQuestion. If evidence type is the unclear part (might be a report they haven't attached yet, might be pure description), lead with that: "New crash-report email (.mad) or a bug you'll describe?" — its answer usually resolves the product question too, since case 1 (if they do have a report) asks about product internally. If evidence type is clear but the product/project genuinely isn't, ask which project instead. - A concrete symptom with a resolvable product/project — everything else. This is the general path (Phase 1 onward). If a product word is named (
bionix,myproduct), read that product's block fromc:\Users\<you>\.claude\skills\light-bug-MadShi\products.ini— you needSourceRoot,ProjectFile,ProjectClaude,BugProtocol,FixesLog,BugFolderRooteven though nothing crashed. A product word alone does NOT imply a crash: a.madonly exists after one, and most bugs reach you as prose, not a report. If no product word is named but a project is obvious from the working directory, use that project's ownCLAUDE.md/ Fixes Log if it has one — proceed with whatever it has, and skip the Fixes Log bookkeeping in Phase 6 if it has none.
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.
- today Changed · +3 lines · -3 tokens per session 4b11fd934ea6
- 4d ago Changed · +10 lines 3de3072e1575
- 8d ago First seen · 120 lines · 206 tokens per session scan A 84ef6ad1ca79
light-bug is a skill published in the GitHub repository GabrielOnDelphi/Claude-Tools-for-Delphi (17 stars, last pushed yesterday), licensed MPL-2.0. It adds 203 tokens to every session and 3,108 once invoked, about $0.0010 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.
Other skills, from other repositories
debug-optimize-lcp
Guides debugging and optimizing Largest Contentful Paint (LCP) using Chrome DevTools MCP tools. Use this skill whenever the user asks about LCP performance, slow page loads, Core Web Vitals optimization, or wants to understand why their page's main content takes too long to appear. Also use when the user mentions…
systematic-debugging
Use when debugging a failing test, build error, or runtime issue that isn't immediately obvious. Guides a 4-phase root cause analysis instead of random fix attempts.
diagnose
Trace from a reproduced symptom to the source code that causes it. Pin the specific file and approximate line, rate confidence in the cause and clarity of the fix independently, and always propose a concrete fix.
repro-admin
Reproduce an EmDash admin UI bug. Attach a container, start the demo dev server, drive the admin with agent-browser using the dev-bypass session, and capture the reproduction as screenshots plus a replayable transcript.
log-error-digest
Analyze log files to troubleshoot errors, identify peak error periods, and produce error clustering, frequency statistics, and time distribution reports. Supports JSON, syslog, and Nginx formats with automatic detection. Use when a user uploads a .log file and asks to analyze errors, find patterns, debug issues, or…
byted-util-volcengine-detect-retry
An orchestration workflow for Volcengine Cloud Detect, a service that checks websites or network endpoints from test locations.