agentsclimarketplace

Bug capture

Skill rohitg00/pro-workflow/skills/bug-capture

Capture a user-reported defect as a durable GitHub issue written in the project's own domain language. Explores the codebase in parallel for context but never leaks file paths or line numbers into the issue. Use when the user reports a bug conversationally, runs a QA pass, or says "file an issue", "log this as a bug", "capture this".From its SKILL.md

Install
npx -y skills add rohitg00/pro-workflow --skill bug-capture

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.

SKILL.md

3.4 KB, 718 tokens by cl100k_base, as published. Nobody here has run it

bug-capture

Turn a conversation into an issue that still reads correctly after a major refactor.

Flow

1. Listen, then clarify minimally

Let the user describe the problem in their own words. Ask at most two short clarifying questions, drawn from:

  • Expected behavior vs. actual behavior.
  • Concrete reproduction steps if not already implied.
  • Frequency: deterministic, intermittent, or one-off.

If the description already answers these, skip straight to filing. Over- interviewing is a tax the reporter pays for your uncertainty.

2. Explore in parallel

While the user is answering, start a background exploration of the relevant area. The goal is not to propose a fix. The goal is to absorb the project's own vocabulary — the nouns and verbs the codebase uses for this feature — so the issue reads like it was written by a maintainer.

If the repo has a glossary file (common names: GLOSSARY.md, UBIQUITOUS_LANGUAGE.md, docs/domain.md), read it first.

3. Check for duplicates before filing

Run gh issue list --search "<key phrase>" --state all --limit 10. If a live or recently closed issue matches, surface it to the user and ask whether to add a comment instead of opening a new issue. Do not silently skip filing.

4. Decide: single issue or breakdown

Break down when the report contains two or more independent failure modes that a different contributor could fix in parallel. Keep as one when every symptom traces to a single wrong behavior.

For a breakdown, file in dependency order so each child issue can reference a real parent number, and mark honest Blocked by links. Avoid inventing dependencies to make the tree look tidier.

5. File with gh issue create

File without asking the user to review the draft. Send back the URLs.

Single-issue template

## What happened
<observed behavior, in domain terms>

## What I expected
<expected behavior>

## Reproduction
1. <step>
2. <step>
3. <step>

## Context
<anything that narrows where the bug lives, in domain terms — e.g.
"only affects the import path, not the export path">

Child-issue template

## Parent
#<parent-number>

## What is wrong
<one behavior, narrow slice>

## What I expected
<expected for this slice>

## Reproduction
1. <step>

## Blocked by
<#issue or "none — independent">

## Context
<notes that apply only to this slice>

6. Rules that apply to every issue body

  • No file paths, line numbers, function names, or PR numbers. These go stale. Describe behavior, not code.
  • Use the project's domain nouns, not generic tech terms. "The sync worker drops the patch" beats "applyPatch() throws".
  • Reproduction steps are mandatory. If you cannot derive them, go back to the user once before filing.
  • Thirty-second read target. Cut anything that does not help a maintainer decide whether to pick it up.

7. Keep going

After each issue, print the URL and ask whether there is a next one. Do not batch multiple reports into one filing pass — each bug deserves its own scoped issue.

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Gives 0 of the 12 instructions most bug reports skills give in 718 tokens

Counted across 130 of the 155 authors here whose files we hold, read 2026-09-06

  • Reproduce the bug as a new failing regression test firstin 11 of 130, across 4 files
  • Add security-reviewer for security-sensitive defectsin 10 of 130, across 3 files
  • Skip the research phase by defaultin 10 of 130, across 3 files
  • Escalate build breaks to build-error-resolverin 10 of 130, across 3 files
  • Stop at Gate 1 and Gate 2in 10 of 130, across 3 files
  • Fix until that test goes greenin 10 of 130, across 3 files
  • Run the orch-pipeline engine with the settings abovein 10 of 130, across 3 files
  • Scope unclear root causes with code-explorer before writing the red testin 10 of 130, across 3 files
  • Replace sensitive data with placeholdersin 6 of 130, across 3 files
  • Match severity rating to actual impactin 6 of 130, across 3 files
  • Reference screenshots and attachments inlinein 6 of 130, across 3 files
  • Create issues as markdown filesin 5 of 130, across 2 files

Said here and by no other author read

  • Let the user describe the problem in their words
  • Explore the codebase to absorb domain vocabulary
  • Read the project glossary first
  • Break down reports with independent failure modes
  • File child issues in dependency order
  • File without asking the user to review drafts

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 325,949. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.