- Home
- Skills
- Testing & Debugging
- Debugging Guide
Debugging Guide
An evidence-based debugging assistant that tracks hypotheses and isolates root causes in web applications.
Free
Works with the AI Agents you already use
Debugging Guide
Example session with this skill installed
My React app's login works locally but fails on staging with a 401. I've checked the API keys. Help me debug this.
- Read your context and instructions
- Compiled the debugging guide
I have reviewed the reported 401 error on staging versus the local success. Since API keys are verified, we need to distinguish between session configuration and environment-specific headers. ## Investigation Status | ID | Possible Cause | Supporting Evidence | Conflicting Evidence | Next Check | Status | |:---|:---|:---|:---|:---|:---| | H1 | Missing CORS/Origin | 401 on cross-origin request | Works locally (same-origin) | Check `Access-Control-Allow-Origin` | Open | | H2 | Secure Cookie mismatch | Staging uses HTTPS | Local uses HTTP | Inspect cookie `Secure` flag | Supported | | H3 | Path/Domain mismatch | 401 implies no session | API keys verified | Check `Set-Cookie` header path | Open | The evidence shows that the request reaches the server but fails authentication. We must determine if the session cookie is being sent by the browser or rejected due to attributes. **Next Check (Proposed)** Inspect the Network tab in your browser for the failed staging request. Look at the `Request Headers` and tell me: 1. Is a `Cookie` header present in the request? 2. In the `Response Headers` of the previous login attempt, what are the `SameSite` and `Secure` attributes of the `Set-Cookie` header? This will distinguish between a failure to store the session (H2) and a failure to send it due to origin policy (H1). **Next steps** - Provide the `Set-Cookie` attributes from the staging login response. - Confirm if the staging API is on a different subdomain than the frontend. - Share a redacted snippet of your session configuration code.
Connects securely to your tools. The creator never sees your data.
About this skill
When a web application breaks, the default response is often a cycle of guessing and "shotgun debugging" that leads to bloated code and phantom fixes. This skill enforces an evidence-based investigation workflow to isolate the root cause without touching production or risking data loss. ## What it does - **Symptom isolation** captures expected versus actual behavior and reproduction steps to define the problem scope. - **Evidence tracking** separates code-backed facts and logs from user observations to avoid bias. - **Hypothesis management** maintains a register of possible causes, supporting evidence, and conflicting data points. - **Diagnostic selection** proposes the lowest-impact check to distinguish between competing mechanisms. - **Verification planning** provides minimal, targeted fixes with specific regression test instructions. ## How it works 1. **Establishment**: The agent queries the current environment, recent changes, and reproduction steps. 2. **Analysis**: It reads entry points and handlers to map the execution path and identify potential failure points. 3. **Investigation**: It presents a hypothesis table and asks the user to perform a specific, non-destructive check. 4. **Resolution**: Once evidence isolates a cause, it proposes a minimal fix and a verification method. ## Frameworks & tools This skill is framework-agnostic and works with any web stack. It relies on standard diagnostic artifacts like status codes, header presence, error logs, and stack traces. ## Why this beats prompting it yourself Generic prompts often jump to "plausible" solutions that do not actually match the evidence. This skill uses a structured investigation guide to prevent "reproduction by assertion" and ensures that every proposed fix is backed by a confirmed, isolated cause. ## Use cases - Debugging intermittent session or authentication failures across different environments. - Investigating silent failures where a component runs but produces no output or incorrect state. - Conducting a handoff-ready investigation for complex bugs that require team review. - Identifying why a specific route or event wiring is failing despite "correct" logic. ## Known limitations Does not perform production mutations, database resets, or deployments. Requires the user to provide sanitized logs and snippets; it cannot autonomously access restricted or secret files.
No production mutations or DB resets. Requires user-provided sanitized logs. Cannot bypass file access restrictions or handle secrets.
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
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 11 days ago
- Free with an Agensi account