- Home
- Skills
- Testing & Debugging
- Regression Triage — What Changed To Break This
Works with the AI Agents you already use
Regression Triage — What Changed To Break This
Find the specific change that broke your code using a systematic, read-only triage workflow and impact analysis.
$12
Regression Triage — What Changed To Break This
Example session with this skill installed
The checkout service started returning 500 errors after the deployment at 2 PM today. It seems to be related to tax calculations. Find what changed.
- Read your context and instructions
- Compiled the regression triage
I have analyzed the history of the src/services/checkout and src/lib/tax modules relative to the 2 PM deployment window.
Regression Triage Report: Checkout 500 Errors
Symptom: HTTP 500 at /api/checkout
Time Window: Post-2 PM Deploy (Commits after 8f2a1c)
Identified Area: src/services/checkout, src/lib/tax
1. Leading Candidates
| Rank | Change | Path | Evidence | Disconfirming Test |
|---|---|---|---|---|
| 1 | a7b3d2 Refactor tax provider interface | src/lib/tax -> src/services/checkout | Architecture signal detected; modifies calculateTax() signature. | Call calculateTax() with legacy arguments; if it passes, this is not the cause. |
| 2 | c9e1f4 Update checkout validation | src/services/checkout | Direct hit on the checkout module; change in validateRequest(). | Mock valid request to validateRequest(); if it returns true, this is not the cause. |
2. Ruled Out
- Commit
b5d6e7(Update README): No code impact.
Commit e3f4g5 (Analytics update): Dependency graph shows no path from src/analytics to src/services/checkout.
3. Analysis Gaps
- I cannot see environment variable changes in the CI/CD pipeline that occurred at 2 PM.
- I have not reviewed the database migration logs which may have occurred during deployment.
Next steps
- Run the disconfirming test for
a7b3d2by checking thetaxmodule's exports. - Verify if the
TAX_API_KEYwas rotated during the 2 PM deployment window. - Check the application logs for a stack trace to confirm if the 500 originates in
calculateTax.
Connects securely to your tools. The creator never sees your data.
About this skill
Debugging a regression often feels like a guessing game when you do not know what moved. This skill systematically works backward from a symptom to the change that caused it, delivering a ranked triage report without ever touching your working tree.
What it does
- Pinpoints symptoms by extracting identifiers like module paths, function names, and error codes from logs or reports.
- Maps impact sets to determine if a specific change can actually reach the reported failure point.
- Ranks candidates based on architecture signals, commit timing, and the size of the impact set.
- Validates hypotheses by generating a specific disconfirming test for every potential cause identified.
How it works
- Frame the symptom by recording the exact error text and the deployment time window.
- Extract identifiers to locate the relevant code area and search the repository for matching modules.
- Establish history using
gitor theagent-contextCLI to find recent commits that modified the target area. - Connect changes by analyzing the dependency graph to see if the modified code impacts the symptom surface.
- Produce the report following a structured template that includes ruled-out suspects and next actions.
Frameworks & tools
This tool uses git for history analysis and optionally integrates with the ai-agent-context CLI for advanced architecture signals and dependency mapping. It works with any codebase where version history is available via Git.
Why this beats prompting it yourself
Generic prompts often lead to "hallucinated" bug fixes or surface-level code reviews. This skill enforces a read-only triage methodology that separates what changed from what broke, ensuring you get evidence-based hypotheses instead of guesses.
Use cases
- Identify which commit broke a specific unit test after a large merge.
- Trace a production timeout back to a specific architectural change or refactor.
- Determine if a new bug is a regression or a long-standing latent issue.
- Verify if a specific module change could have caused a failure in a distant consumer.
Known limitations
This is a read-only triage tool that does not reproduce failures or apply fixes. It is limited to the most recent repository history and cannot analyze changes outside the current repository.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Testing & Debugging.
Read-only analysis only. Does not reproduce bugs or apply fixes. Recent history is limited to the last 20 commits when using the optional CLI. Cannot access remote logs or external CI data.
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 buying a skill.
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
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 28 days ago
- Software
- ai-agent-context
Node.jsagent-context
Git