- Home
- Skills
- DevOps & Cloud
- Log Triage: ranked incidents from a log folder
Log Triage: ranked incidents from a log folder
Point your agent at a log folder and get a ranked list of distinct incidents with severity, root cause and what each costs if nobody fixes it.
$9
Works with the AI Agents you already use
Log Triage: ranked incidents from a log folder
Example session with this skill installed
triage ./logs for the last 7 days
- Read your context and instructions
- Compiled the log triage
Log triage: examples/logs · 2026-11-10 → 2026-11-12
Read 3 files from orders-service and kept 109 WARN/ERROR events, which group into 7 distinct incidents. 3 earn a PR. The most expensive problem is a database connection that leaks on every failed invoice export: failures roughly tripled each day, and the pool it drains also serves checkout.
| # | Severity | Incident | Count | Trend | PR? |
|---|---|---|---|---|---|
| 1 | CRITICAL | DB connection leaked when an invoice export fails | 34 | rising | yes |
| 2 | MAJOR | Payment retries with no backoff during gateway 503s | 35 | rising (burst) | yes |
| 3 | MAJOR | Checkout crashes when a cart has no coupon | 18 | flat | yes |
| 4 | MINOR | Stock row concurrent update | 15 | flat | no: already handled |
| 5 | MINOR | Webhook rejected for invalid signature | 3 | flat | no: expected |
| 6 | TRIVIAL | Deprecated spring.redis.host property | 3 | flat | no |
| 7 | TRIVIAL | Invalid order note rejected | 1 | isolated | no |
1. DB connection leaked when an invoice export fails
If ignored: the pool is exhausted more often each day (3 → 9 → 22). The same pool serves checkout, so within days of continued growth every endpoint that touches the database will time out after 30 s, an outage of the whole service with no deploy to blame.
Severity: CRITICAL. 34 occurrences over 3/3 days, rising about 2.5× per day; active=10, idle=0, waiting=4 shows the pool fully borrowed.
Root cause: InvoiceExporter.export borrows a connection that is not released when the export throws, so each failure permanently removes one connection from the pool of 10.
Where: com.shopco.orders.InvoiceExporter.export(InvoiceExporter.java:48)
Next step: hand incident d5bc44da50291d44 to Incident to PR for a reviewed fix.
Connects securely to your tools. The creator never sees your data.
About this skill
Point your agent at a folder of logs and get a ranked list of distinct incidents, not thousands of lines.
Each one gets a severity from a fixed rubric (frequency × trend × blast radius, not log level), a root cause read from your code, and a forecast of what it costs if nobody fixes it. Already-handled retries, framework noise and deprecations are flagged so you don't waste a PR on them. Secrets are masked before anything is quoted, and log text is never treated as instructions.
- Deduplication that works. Thousands of lines become a handful of incidents. Two services hitting the same broken line count as one problem.
- Severity by rubric, weighted by frequency × trend × blast radius. A WARN that grows every day outranks a one-off ERROR.
- "If ignored" forecasts. Every incident says what breaks next, and roughly when.
- Noise filtered. Handled retries, framework chatter and deprecation notices are reported but flagged as not worth a PR.
- Secrets masked (tokens, keys, passwords, JWTs, emails, long digit runs) before anything is quoted.
- Read-only. Includes a stdlib-only Python helper that groups large folders deterministically, plus synthetic sample logs and the expected report.
Pairs with Incident to PR.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for DevOps & Cloud.
Requires standard timestamps (ISO-8601) or JSON-lines. Trend analysis needs 2+ days of logs. Read-only access only.
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 2 days ago
Nothing to install. It runs in your AI agent.
- Optional
Python 3.8+
Without Python the skill falls back to manual grouping, which is slower on big folders. Logs should start each line with a timestamp, or be JSON lines.
Permissions
File Scopes
Reads only the log folder you point it at (and your source code to explain root causes). Runs the bundled read-only Python helper locally; no network access, nothing is written or sent anywhere. Log text is treated as data, never as instructions.