- Home
- Skills
- Testing & Debugging
- triage trip
More screenshots
Works with the AI Agents you already use
triage trip
>-Automates the triage of AI-introduced regressions by linking runtime errors with your local git history and AI prompt metadata to generate precise, minimal patch fixes.
Free
triage trip
Example session with this skill installed
The build failed with "ReferenceError: user_id is not defined" in src/services/auth.ts after I let Cursor refactor the login flow. Find the break and fix it.
- Read your context and instructions
- Compiled the triage trip
I have analyzed the git diff and Cursor prompt history. The regression was caused by a prompt at 14:22:10 which renamed the user_id property to userId in the Session interface but failed to update the reference in auth.ts.
Triage Report
| Category | Details |
|---|---|
| Root Cause | src/services/auth.ts line 42 |
| Attributed Prompt | "Refactor session handling to use camelCase for all user properties" |
| The Shift | user_id (deprecated) → userId (current) |
| Action | Update caller to match the new schema |
--- a/src/services/auth.ts
+++ b/src/services/auth.ts
@@ -39,7 +39,7 @@
export const validateSession = (session: Session) => {
- if (!session.user_id) {
+ if (!session.userId) {
throw new Error("Invalid session");
}
Next steps
- Run
git applyto fix the reference error. - Verify the fix by running the build command.
- Review
git diffto ensure no other callers were missed.
Connects securely to your tools. The creator never sees your data.
About this skill
Fixing regressions introduced by AI coding assistants is tedious. When Cursor or Copilot refactors a file, it often breaks API contracts, renames schema fields, or drops function parameters without updating callers. This skill automates the triage process by linking runtime errors directly to the breaking git diff and the specific AI prompt that caused it.
What it does
- Capture errors by parsing runtime stack traces or build failure logs provided by the user.
- Analyze git context by inspecting staged changes, working tree diffs, and the last five commits.
- Trace prompt provenance by querying local Cursor metadata to identify which specific prompt triggered the regression.
- Diagnose shifts by identifying mismatches between old and new contracts, such as
userIdbecominguser_id. - Generate patches in a minimal unified diff format to restore functionality without bloat.
How it works
- Context gathering: The agent runs
git diffand a local script to extract recent Cursor prompts from the SQLite database. - Root cause analysis: It applies a diagnosis rubric to map the error to a specific line of code and a recent change.
- Patch emission: It generates a precise unified diff that aligns callers with the updated schema.
- Application: The agent presents a four-section report and waits for user confirmation before applying the fix via
git apply.
Frameworks & tools
This skill requires a local git repository and Node.js (18+). It utilizes sqlite3 for prompt history retrieval and works with standard unified diff tools.
Why this beats prompting it yourself
Standard prompts often suggest generic fixes or defensive null checks that mask the real issue. This skill systematically correlates your local git history with AI metadata to find the exact "intent shift" that broke the build, ensuring the fix addresses the root cause rather than just the symptom.
Use cases
- Debugging a
TypeErrorimmediately after an AI-driven refactor. - Identifying which specific Cursor Composer prompt broke a TypeScript interface.
- Resolving "variable not defined" errors caused by AI renaming local exports.
- Generating a minimal patch to fix breaking schema changes in a monorepo.
Known limitations
Requires a git repository for diff analysis. Cursor prompt retrieval is best-effort and depends on local sqlite3 availability. Does not replace the need for running a full test suite after patching.
Requires a git repository for diffs. Cursor prompt provenance is best-effort and requires local sqlite3. Native path does not use external APIs. Requires Node 18+ for prompt extraction.
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 20 days ago
- Free with an Agensi account
Requires