---
title: "Prod Logs Health Check: review, role & definition · Undominated.ai"
canonical: https://undominated.ai/agents/wshobson-prod-logs-health-check/
description: "Pulls recent production logs after a deploy or load test and reports errors grouped by root cause, counting distinct failures separately from retries — treating raw logs, not dashboards, as the only acceptable primary source."
---

# Prod Logs Health Check: review, role & definition · Undominated.ai

> Pulls recent production logs after a deploy or load test and reports errors grouped by root cause, counting distinct failures separately from retries — treating raw logs, not dashboards, as the only acceptable primary source.

[← Explore all agents](/agents/)

OPERATIONS AND RELIABILITY / wshobson

# Prod Logs Health Check

Pulls recent production logs after a deploy or load test and reports errors grouped by root cause, counting distinct failures separately from retries — treating raw logs, not dashboards, as the only acceptable primary source.

 [Use this definition ↓](#setup)[Original source ↗](https://github.com/wshobson/agents/blob/4236bb91f8395b0435f1d8b8baf9e8e4c69a8620/plugins/operating-kit/agents/prod-logs-health-check.md)

SOURCE REVIEW

 Reviewed 2026-10-07
 Evidence 3 linked sources
 Publisher wshobson
 Licence [MIT ↗](https://github.com/wshobson/agents/blob/4236bb91f8395b0435f1d8b8baf9e8e4c69a8620/LICENSE)
 Revision 4236bb91f839
 [Read what was—and wasn’t—checked ↓](#review)

“Pulls recent production logs filtered for errors, warnings, and anomalies. Use after any deploy, after a load test, or any time you suspect something is going wrong. Treats logs as the only acceptable primary source for incident analysis — never infers from dashboards or script stdout alone.”

 [wshobson · upstream description ↗](https://github.com/wshobson/agents/blob/4236bb91f8395b0435f1d8b8baf9e8e4c69a8620/plugins/operating-kit/agents/prod-logs-health-check.md) Our analysis follows below.

01 / THE REASONING

## Why this made the selection.

 - Makes raw logs the mandatory primary source and refuses to present inference as fact — if logs were not checked, the finding must say so before anything is presented.
- Cross-references job ids before counting, so one failure retried five times is reported as one distinct failure, not five.

### A good fit for

 - Post-deploy and post-load-test verification against real log lines rather than dashboard summaries.
- First-pass incident triage that groups errors by root cause and states what could not be confirmed.

### Weigh up before choosing

 - The definition is a template: the {{LOG_QUERY}} and {{PROJECT_SPECIFIC_MARKERS}} placeholders must be pointed at the project’s real log source before use.
- Declares only Bash and Read; it has no structured-logging or time-series tooling and cannot paginate long windows itself.
- Selects the haiku model alias; availability and behaviour depend on the host account.

02 / THE REVIEW RECORD

## What we actually inspected.

Source review has boundaries. A clear record is more useful than a “safe” badge.

### Material inspected

 - plugins/operating-kit/agents/prod-logs-health-check.md (complete frontmatter and body at revision 4236bb9)
- LICENSE (MIT redistribution terms at revision 4236bb9)
- YAML frontmatter parsed; sha256 of the exact raw bytes recorded and cross-checked against the git blob at the pinned revision

### Our findings

 - The core rule forbids analysing an incident from UI data or script stdout alone, on the stated grounds that dashboards paginate and test-harness timing misleads for async work.
- The report contract requires the time window, the number of lines pulled (so truncation is visible), grouped root causes with representative excerpts, and explicit open gaps.

### Not established by this review

 - The agent has not been executed or benchmarked.
- Tool availability, host/model compatibility and task outcomes were not runtime-tested.

The review applies to the material and revision named here. A newer upstream release can change its behavior.

03 / PUT IT TO WORK

## Use the role in your project.

[Upstream setup instructions ↗](https://code.claude.com/docs/en/sub-agents.md)
 - Download the original prod-logs-health-check.md with its full licence and attribution; inspect the complete instructions before use.
- Replace the {{LOG_QUERY}} template with the project’s real log source (cloud logging, journald, a file, kubectl logs) and fill in project-specific failure markers.
- Place the working copy at .claude/agents/prod-logs-health-check.md and invoke it after deploys or when an incident is suspected.
- Scope the Bash grant to the log-query command the template needs; the agent reads and reports only.

### Before you start

 - Claude Code with repository read access and a Bash grant scoped to the project’s log source.
- Access to real production logs; the agent reports their absence rather than substituting dashboards.

### Compatibility

Claude Code subagents

### Log retrieval over Bash plus repository reading

 - Declares Bash and Read; the Bash grant runs the project-specific log query, so scope it to read-only log access.
- The agent reports findings only and does not modify the system under investigation.

### Cost model

The definition is reusable under MIT. The host, model and any connected services have their own access and billing terms.

THE COMPLETE REVIEWED DEFINITION

## Read it before you reuse it.

Original source bytes, with attribution. Review the host-specific setup notes above.

 Copy definition ↗ [Download definition + licence ↗](/resources/agents/wshobson-prod-logs-health-check/bundle.zip)[Raw Markdown ↗](/resources/agents/wshobson-prod-logs-health-check/definition.md)
 ---
name: prod-logs-health-check
description: Pulls recent production logs filtered for errors, warnings, and anomalies. Use after any deploy, after a load test, or any time you suspect something is going wrong. Treats logs as the only acceptable primary source for incident analysis — never infers from dashboards or script stdout alone.
model: haiku
tools: Bash, Read
---

You are this project's production-log health checker. Pull real logs and report what's
actually happening, not what a dashboard claims is happening.

**Template note:** point `{{LOG_QUERY}}` at the project's real log source
(cloud logging, journald, a file, `kubectl logs`, etc.).

## Core rule

Never analyze a production incident from UI data or script stdout alone. Dashboards paginate
(you see the last N events, not all), and test harness timing is often wrong for async work.

If logs are not available or you didn't check them, say so explicitly before presenting any
finding. Do not present inference as fact.

## Steps

### 1: Pull recent logs
```bash
{{LOG_QUERY}}
```

### 2: Filter for signal
Grep for:
- Errors, exceptions, stack traces
- Timeouts, retries
- Project-specific failure markers: `{{PROJECT_SPECIFIC_MARKERS}}`

### 3: Distinguish unique failures from retries
The same job id appearing 5 times is one failure retried, not five failures.
Cross-reference ids before reporting a count.

## What to report
- Time window and how many log lines you pulled (so truncation is visible).
- Errors grouped by root cause, with a representative excerpt each.
- Distinct-failure count vs. total occurrences.
- Anything you could not confirm from logs, stated as an open gap.

The download contains prod-logs-health-check.md . Keep its filename when placing it in the agent directory described above.

By **wshobson**. [Exact upstream source ↗](https://raw.githubusercontent.com/wshobson/agents/4236bb91f8395b0435f1d8b8baf9e8e4c69a8620/plugins/operating-kit/agents/prod-logs-health-check.md) · [Licence](/resources/agents/wshobson-prod-logs-health-check/LICENSE.txt) · [Attribution](/resources/agents/wshobson-prod-logs-health-check/ATTRIBUTION.txt)

SHA-256 3ecb76b643dadd8e0a84efc598c3ce7bf55694ea4864e4400920eb2b98af547f

 Read the applicable licence MIT License

Copyright (c) 2024 Seth Hobson

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.

04 / FOLLOW THE EVIDENCE

## The source trail.

Our notes are separate from the original resource. Check upstream before adopting a new version.

 - [Complete upstream definition at the reviewed revision ↗](https://github.com/wshobson/agents/blob/4236bb91f8395b0435f1d8b8baf9e8e4c69a8620/plugins/operating-kit/agents/prod-logs-health-check.md) Checked 2026-10-07 https://github.com/wshobson/agents/blob/4236bb91f8395b0435f1d8b8baf9e8e4c69a8620/plugins/operating-kit/agents/prod-logs-health-check.md Supports: summary, upstreamDescription, whySelected, bestFor, limitations, review, access
- [Applicable full upstream licence ↗](https://github.com/wshobson/agents/blob/4236bb91f8395b0435f1d8b8baf9e8e4c69a8620/LICENSE) Checked 2026-10-07 https://github.com/wshobson/agents/blob/4236bb91f8395b0435f1d8b8baf9e8e4c69a8620/LICENSE Supports: license, artifact
- [Current official custom-agent configuration ↗](https://code.claude.com/docs/en/sub-agents.md) Checked 2026-10-07 https://code.claude.com/docs/en/sub-agents.md Supports: compatibility, install, access, limitations, review

KEEP COMPARING

## Other approaches to consider.

Related by category or shared topics. These are alternatives to inspect, not a measured quality order.

 [### New Relic Incident Response Agent ↗ Assesses a New Relic alert, investigates traces, errors and deployments, and proposes a code fix only after the engineer agrees to each phase.](/agents/github-new-relic-incident-response/)[### PagerDuty Incident Responder ↗ Pulls a PagerDuty incident, correlates it with commits from the 24 hours before it started, and proposes a fix or rollback pull request.](/agents/github-pagerduty-incident-responder/)[### Performance Monitor ↗ Reads existing metric, log and output files to surface performance anomalies, then writes an evidence-cited observability plan — every reported number must have been read from a file or counted, never invented.](/agents/voltagent-performance-monitor/)

 [AI Tools ↗](/tools/)[Skills ↗](/skills/)[Agents ↗](/agents/)[MCP Servers ↗](/mcp-servers/)[Workflows ↗](/workflows/)

## Continue your investigation

 - [Find reusable instructions](/skills/)
- [Inspect connections](/mcp-servers/)
- [Build a toolkit](/tools/)
- [Choose the model](/compare/)
