- Home
- Skills
- Agents & Orchestration
- Systematic Debugging
Works with the AI Agents you already use
Systematic Debugging
A disciplined debugging methodology that enforces root cause isolation and regression testing before applying fixes.
$10
Systematic Debugging
Example session with this skill installed
Debug the 'order-total-mismatch' symptom. Repro: 'npm test tests/calc.test.js'. Sources: 'src/calc.js', 'src/tax.js'. Test path: 'tests/regression.test.js'. Goal: Fix rounding error in totals.
- Read your context and instructions
- Compiled the systematic debugging
I have completed the systematic debugging cycle for the order-total-mismatch symptom. The root cause was isolated to a floating-point precision error during tax calculation, which has been resolved and verified. ## Debug Run Record: Order Processor | Field | Value | | :--- | :--- | | **Product** | Order Processor | | **Slice Goal** | Fix rounding error in totals | | **Symptom** | order-total-mismatch | | **Status** | ready | ### Reproduce Command `npm test tests/calc.test.js` failed as expected. **Failure:** `Expected total 100.00 but got 99.9999999998` ### Isolate The error persists even with simplified 10% tax rates. Tracing `src/tax.js` showed the multiplier was applied before rounding. **Root cause is** the tax calculation utility returns raw floats to the aggregator instead of rounding to the smallest currency unit because the `calculateTax` function lacked a precision floor. ### RED Created `tests/regression.test.js`. Command `npm test tests/regression.test.js` failed with the precision mismatch. ### GREEN Applied fix to `src/tax.js` using `Math.round`. 1. `npm test tests/regression.test.js` - **PASS** 2. `npm test tests/calc.test.js` - **PASS** 3. `npm test` (full suite) - **PASS** **Next steps** - Review the `plans/debug-run.md` file for the full audit trail. - Merge the changes in `src/tax.js` and the new regression test. - Start a new cycle if there are other symptoms in the backlog.
Connects securely to your tools. The creator never sees your data.
About this skill
The problem
Debugging by trial and error leads to "shotgun patching" where code is modified without understanding the root cause. This results in brittle fixes, cluttered codebases, and regressions that go unnoticed until they hit production.
What it does
- Enforces a strict reproduce-isolate-fix-verify cycle for a single specific symptom.
- Validates that the reproduction command actually fails before any code analysis begins.
- Isolates the root cause into a single verifiable sentence supported by evidence.
- Requires a failing regression test to be established before the fix is applied.
- Verifies the fix against the reproduction loop, the new regression test, and the existing suite.
Frameworks & tools
This skill is framework-agnostic and works with any existing language, test runner, or shell-based build system. It uses standard file operations and shell commands provided by the environment.
Why this beats prompting it yourself
Standard LLM prompts often lead to premature patching or guessing. This skill forces a disciplined engineering method that forbids production changes until a root cause is proved, ensuring high-integrity fixes and automated documentation of the debug journey.
Use cases
- Root cause analysis for complex logic errors in backend services.
- Fixing intermittent failures by narrowing down load-bearing environment factors.
- Resolving regressions by isolating the specific change that introduced a bug.
- Documenting critical bug fixes for audit trails and technical debt reviews.
Known limitations
Only handles one symptom per cycle. It will not install new test frameworks or handle highly flaky bugs (repro rate under 50 percent).
Processes one symptom at a time. Requires existing test runners. Stops if a reproduction command cannot be established or if three fix attempts fail.
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
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 2 months ago