Bug Report Your Agent Can Fix
by PaperChaseWebb
A no-install way to turn "it is broken" into a report a coding agent can act on: a capture checklist, one report template, a redaction step for logs, a read-only checker, and four prompts that make the agent reproduce, fix, test and write back what changed.
- Price
- $14
- Difficulty
- Beginner
- Setup
- 10 min
- Format
- .md file
bug-report-your-agent-can-fix.md · 17 KB · instant download after checkout
Want this built for you instead? We can build it for you.
What you end up with
Bug reports arrive with steps, evidence and a done-when line, with secrets masked, and the agent proves the fix with a test before you review it.
What you get
- A seven-step capture checklist for whoever finds the bug
- An eight-section report template
- redact.py: masks emails, tokens, keys, cookies and passwords in logs
- bug_check.py: 16 read-only checks on a report and its attachments
- Four prompts: reproduce, find the cause, fix and test, write back
- A worked example and 13 tests
Who it is for
Anyone who hands bugs to a coding agent and wants reports it can reproduce, with nothing private in them.
Works with
Solid chips: the agent is named in this listing's own tools list. Dashed chip: Plain Markdown, so it also pastes into other agents such as Grok Bot, Hermes, OpenClaw or Muse. We list those as compatible by format only, not as tested.
Tools needed
Cost to run: The pack is the only purchase. The scripts run locally with no API calls.
Time to result: A first checked report in about 10 minutes
How it works
Capture
Whoever finds the bug follows the checklist in section 3 and fills in report-template.md.
Expected: A report with numbered steps, expected, actual, environment and evidence.
Redact and check
Run python3 redact.py on each log, then python3 bug_check.py report.md --attachments evidence/.
Expected: 0 findings.
Hand it to the agent
Paste prompts 1 to 4 in order.
Expected: A failing test that then passes, and a Fix note under the report.
Sample prompts, free to use
Read report.md and the files in evidence/. Do not change any code yet. Reproduce the bug by following the steps exactly, or write a failing test that shows it. Tell me the command you ran and its output. If you can not reproduce it, say which step differs and stop.
Prompt 1 of 4.
Write a short note under the report titled Fix: the cause in one sentence, the files changed, the test that proves the Done when line, the commands you ran with their results, and anything you did not test. Do not deploy or merge; I will review first.
Prompt 4 of 4.
Free preview of the file
# Bug Report Your Agent Can Fix A no-install way to turn "it is broken" into a report Claude Code, Codex or Cursor can act on: a capture checklist for whoever finds the bug, one report template, a redaction step before anything leaves the machine, a read-only checker, and the prompts that make the agent reproduce, fix, test and write back what changed. Version 1.0, 2026-10-30. Plain Markdown plus two Python 3 scripts, standard library only, no network. ## 1. What you get - A capture checklist for the person who finds the bug (section 3) - The report template, eight sections (section 4) - redact.py: masks emails, bearer tokens, API keys, cookies and passwords in logs (section 6) - bug_check.py: sixteen read-only checks on a report and its attachments (section 7) - Four prompts: reproduce first, fix, test, write back (section 8) - A worked example and 13 tests (sections 9 and 10) ## 2. Unpack Save this file as pack.md in an empty folder. Each script below starts with a "File:" line. Copy each block into a file with that name, or ask your agent: "Read pack.md and write every File: block to disk with that exact name. Do not run anything." ## 3. The capture checklist (for whoever finds the bug) Do these before you close the tab or restart the app. 1. Write down the time, with time zone. 2. Take a screenshot of the broken state. A short screen recording is better if it moves. 3. Open the browser console (or the app log) and copy the lines around the first error, about ten
Test record
- Date
- Oct 30, 2026
- What was run
- tests/test_bug.py (13) on the template, a worked example and a raw console log with planted secrets. Linux, Python 3.13.
- Result
- 13 passed, 0 failed; both scripts left the folder byte-identical; redact.py refused to overwrite; exit 2 on missing inputs.
- Not run
- The prompts inside Claude Code, Codex or Cursor. Real production logs or HAR files. Images.
Copied from this item's dated test notes. What was not run stays in.
Worth knowing
- redact.py masks what its patterns match. It can not see inside screenshots or recordings, and unusual secret formats may slip through; read the redacted copy before sharing.
- The prompts were not run inside Claude Code, Codex or Cursor.
- On 2026-10-30, on Linux with Python 3.13: 13 tests passed. The first run found redacted values still tripped two checks; the patterns were fixed and the test kept.
Questions
Do I need to install anything?
No. The template and prompts are Markdown. The two scripts need only Python 3.
Does it send my logs anywhere?
No. Both scripts run locally with no network, and redact.py never edits the original file.
Details
Version 1.0 · Updated Oct 30, 2026 · Markdown version of this page
License: Use inside your own team or company, including on client work. Do not resell or republish the pack itself. Download terms
Related
Bug Report Your Agent Can Fix (free sample)
From the $14 pack: the capture checklist, the report template, five rules and the reproduce-first prompt.
Production Readiness Audit: decide from evidence whether an agent system can go live
Twelve areas, each pass, gap or na, and every pass points to a dated evidence file. Gaps get a named owner and a due date, five areas block go-live until fixed, drills are run rather than described, and the verdict is signed by someone who did not build the system. A fifteen-check script reads the folder, the report is sealed to a hash of the evidence, and the owner's go-live yes is tied to the report's sha256.
Extension Lockfile: pin your editor extensions and check for persistence after a bad update
Write down the editor extensions you trust at the versions you reviewed, turn off silent updates, and check every machine against that list with a read-only script. Take a baseline of SSH keys, startup entries, crontab and shell files on a quiet day, so when the next bad extension update is announced you can see what changed in one command. Includes an indicator file and runbook for the 2026-10-29 Claude Code extension incident.