- Home
- Skills
- Testing & Debugging
- Bug Reproduction Receipt — CI & Flaky Tests
Works with the AI Agents you already use
Bug Reproduction Receipt — CI & Flaky Tests
Turn a bug report into a failure another engineer can rerun, then check the same case after a requested fix.
Free
Bug Reproduction Receipt — CI & Flaky Tests
Example session with this skill installed
Synthetic local example. The supplied slug function should reject empty and whitespace-only titles. It currently checks emptiness before trimming:
def slug(value):
if not value:
raise ValueError("title is empty")
return "-".join(value.strip().lower().split())
Reproduce the whitespace-only failure using check_slug.py. The check expects both "" and " " to raise ValueError, while "Local Events" must produce "local-events".
After reproducing it, fix only slugger.py and rerun the unchanged check. Keep two before and two after attempts. Starting fixture revision: 2321739e43668190c456ceeeef6767dc3251044c.
- Read your context and instructions
- Compiled the bug reproduction receipt
Whitespace failure reproduced; fix verified on this case
Actual synthetic local run on macOS with Python 3.14.3. The same python3 check_slug.py case ran twice before and twice after the requested change.
| Observation | Before | After |
|---|---|---|
| Empty title | Rejected as expected | Rejected as expected |
| Three spaces | Returned ''; expected ValueError | Rejected as expected |
Local Events | Returned local-events | Returned local-events |
| Attempts / failing attempts | 2 / 2 | 2 / 0 |
| Exit codes | 1, 1 | 0, 0 |
Cause: the original guard checked the untrimmed value. Three spaces passed it, then became an empty slug.
Only slugger.py changed. The corrected function trims first:
def slug(value):
cleaned = value.strip()
if not cleaned:
raise ValueError("title is empty")
return "-".join(cleaned.lower().split())
The reproducer's bytes were unchanged across all four attempts; its SHA-256 is aa9665305caf9b98faf4d3ed892382c89894a3a43da766b661c4db46d9585d53.
Handoff: run the same check in the target project environment before accepting the patch there. This deterministic fixture was not a CI run, a production application test or a measurement of flaky-failure rates.
Connects securely to your tools. The creator never sees your data.
About this skill
A bug report says something broke. A patch says it is fixed. Keep a runnable case that connects the two.
Bug Reproduction Receipt helps your agent reduce the report to the input, setup and command that expose the failure. It records the revision, environment, expected result and actual signal so the next engineer can repeat the observation.
Keep the same test before and after
When you request a fix, the workflow follows the reproduced path, makes the smallest supported change and reruns that same case with relevant adjacent checks. For an intermittent failure, it records bounded attempts and failures on both sides of the change.
You receive a reproduction recipe and separate conclusions: whether the original failure was reproduced, and whether a requested fix was verified on that case. If execution is unavailable, the recipe remains useful and its result stays unverified.
Use it for a bug handoff, a failed CI investigation or a disputed fix. It runs through your agent and project tools, with no hosted service or MCP server required. A short successful run does not prove an intermittent bug can never return.
A failure may depend on unavailable infrastructure, private data, timing, or environment. Bounded attempts do not prove flakiness eliminated. A code-path hypothesis without execution is not a reproduced runtime failure.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
1 use
Times used to date
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 25 days ago
- Free with an Agensi account
Permissions
Read project files and run local tests; write files only when a fix is requested. Browser access only for browser-visible defects. No default external account, credential, or hosted-service access.