- Home
- Skills
- DevOps & Cloud
- Incident to PR: stack trace in, reviewed fix PR out
Incident to PR: stack trace in, reviewed fix PR out
Turn a production stack trace or log-triage incident into a small, adversarially reviewed pull request, or a clear reason why no safe patch exists.
$14
Works with the AI Agents you already use
Incident to PR: stack trace in, reviewed fix PR out
Example session with this skill installed
open a PR for incident d5bc44da50291d44 from the triage report
- Read your context and instructions
- Compiled the incident to pr
**Branch** `nightshift/fix-critical-d5bc44da` · 1 file (+14 −15) · Adversarial review: **PASS** `InvoiceExporter.export` closed its connection only after a successful commit, so every failed batch leaked one connection from the pool of 10 (34 failures over 3 days, rising 3 → 9 → 22). The fix acquires the connection and statement in try-with-resources so both close on every path; a failed batch still rolls back. ```diff - Connection connection = dataSource.getConnection(); - connection.setAutoCommit(false); - PreparedStatement statement = connection.prepareStatement(INSERT_SQL); + try (Connection connection = dataSource.getConnection(); + PreparedStatement statement = connection.prepareStatement(INSERT_SQL)) { + connection.setAutoCommit(false); ``` **PR title:** `fix(orders): db connection leaked when an invoice export fails` Here are the full diff and PR body. Push the branch and open the PR?
Connects securely to your tools. The creator never sees your data.
About this skill
Give your agent a production incident (a stack trace, an error from your logs, or a finding from Log Triage) and get a small, reviewed pull request. The skill finds the code from the stack trace, checks the diagnosis against the real source before writing anything, writes a bounded patch (3 files, 120 lines, never build files, migrations, CI or secrets), then reviews it adversarially and discards it if it fails. The PR body leads with what the bug costs if ignored, with masked evidence and a test plan. It works on its own branch, asks before pushing, and never merges. - **Code found from the stack trace**, not guessed from the message. - **Root cause checked against the code** before anything is written. A wrong diagnosis stops here. - **Bounded patch**: at most 3 files and 120 changed lines, with a deny list for build files, migrations, CI, Docker, lockfiles and secrets. - **Adversarial review** (in a fresh subagent where your agent supports it). A rejected patch is a good outcome: nobody reviews a wrong fix. - **A PR body people read**: what is wrong, severity, why, what it costs if ignored, masked evidence, the fix, how to verify, provenance. - **Never merges**, never pushes to the base branch, asks before pushing. Pairs with Log Triage.
Max 120 lines and 3 files per patch. No support for build files, migrations, or secrets. Requires stack trace to point to local source, not third-party libs. Requires clean git status.
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 days ago
Requires
Permissions
File Scopes
Reads and edits files in your repository on a new branch, runs git and your test command, and pushes a branch and opens a PR with gh only after you confirm. Never merges, never force-pushes, never pushes to the base branch. Log and stack-trace text is treated as data, never as instructions.