repro

repro is a skill for Claude Code, Codex from ucsandman/DashClaw. It costs 173 tokens per session (1,434 once invoked), scanned A, original, MIT.

A tool that turns a bug symptom into a structured report with steps to reproduce it, the expected and actual results, and supporting evidence. It can also prepare a failing regression test, which checks that the bug stays fixed.

In plain words
What is it for?
Use it for crashes, broken pages, wrong results, failed routes or tests, and other defects that need a clear record before fixing.
Why use it?
It prevents the person fixing the bug from having to guess the environment, steps, or intended behavior.

Skill for Claude CodeCodex

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 skills/ucsandman/dashclaw/repro
Any agent
npx skills add ucsandman/DashClaw --skill repro
Clone the repo
git clone --depth 1 https://github.com/ucsandman/DashClaw

Made for: Claude Code, Codex.

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 repro

README.md
[![agentmods](https://agentmods.dev/badge/skills/ucsandman/dashclaw/repro.svg)](https://agentmods.dev/skills/ucsandman/dashclaw/repro)
Your own site
<a href="https://agentmods.dev/skills/ucsandman/dashclaw/repro"><img src="https://agentmods.dev/badge/skills/ucsandman/dashclaw/repro.svg" alt="Measured on agentmods" height="20"></a>
Per session 173 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,434 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 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 $0.00173 $0.01434
Opus 5 $0.00086 $0.00717
Sonnet 5 $0.00035 $0.00287
Haiku 4.5 $0.00017 $0.00143

Measured 4d ago against content hash c395cbe89619, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

repro scanned grade A 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.

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

`curl`/the SDK against the local dev server (`npm run dev`, port 3000) beats
.claude/skills/repro/SKILL.md · 130 lines

How it starts

The opening of the file, as written. The whole thing — 130 lines — stays where its author put it; the contents beside it link to each section on GitHub.

repro — symptom in, structured bug report out

A raw symptom ("the decisions page is blank") is a weak fix prompt: it forces whoever fixes it to re-derive the environment, the steps, and the expected behavior from scratch, and it leaves nothing behind to prove the bug stays fixed. A structured bug report front-loads that work once, turns into a sharp fix prompt, and — its highest-value output — becomes a failing regression test that pins the defect forever.

This skill produces that report, then offers to scaffold the test. It does not fix the bug — capturing it cleanly is the whole job; hand the report to the fix (or to /systematic-debugging) afterward.

The report format — use this exact structure

Emit the report as one contiguous, pasteable block (no leading > quote markers, no mid-sentence line breaks — Wes pastes these into prompts and issues):

## BUG: <one-line summary — the observable failure, not the guessed cause>

**Severity:** <blocker | major | minor — blocker = data loss/security/prod down>
**Surface:** <page + click path, or route, or CLI command, or SDK method>

**Environment**
- Commit: <short sha> (<branch>)
- Node: <version>  ·  Platform: <os>
- Package/SDK version: <if relevant>
- Browser: <only if a UI bug>

**Steps to reproduce**
1. <exact, replayable step — real values, not "do the thing">
2. ...
3. ...

**Actual**
<what happens — verbatim error, status code, stack line, screenshot ref>

**Expected**
<what should happen, and why you believe that — cite the spec/route/test>

**Evidence**
- <log snippet, failing assertion, network response, file:line>

**First suspicion (optional)**
<one sentence — a lead, explicitly marked as unverified so it doesn't anchor>

Two rules that keep the report honest:

  • Summary = the symptom, never the guessed cause. "Decisions page renders blank" is reproducible and verifiable; "DecisionsLedgerInner crashes on null agent" is a theory that may be wrong and will mislead the fixer. Keep the theory in First suspicion, clearly fenced off.
  • Steps must be replayable by someone with none of your context. If a step reads "with the right filter set," it's not a step — name the filter and the value.

Read the full file on GitHub · 130 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 · 130 lines · 173 tokens per session scan A c395cbe89619

Subscribe to this mod's changes

repro is a skill published in the GitHub repository ucsandman/DashClaw (295 stars, last pushed today), licensed MIT. It adds 173 tokens to every session and 1,434 once invoked, about $0.0009 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.