mcpField
S
solarwinds-logs

solarwinds-logs

github.com/jakenutsCloud & DevOps2113 installsUpdated 16w ago
Repository

About solarwinds-logs

solarwinds-logs is a skill for searching DealerVision production logs through the SolarWinds Observability API. It uses the `logs` CLI and outlines how to filter, inspect, page through, and export results.

Key features

  1. 01Search production logs by keyword
  2. 02Filter by severity, service, or host
  3. 03Inspect entries and structured data
  4. 04Page through results and export files
  5. 05Guidance on using time arguments

Use cases

  1. 01Investigate production errors and exceptions
  2. 02Locate timeout logs by service or host
  3. 03Inspect details for a specific log entry
  4. 04Export large log sets for analysis

MCP support

The README refers to documentation for MCP server modes but does not describe how this skill uses MCP.

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/jakenuts/agent-skills --skill solarwinds-logs

After installing, type/solarwinds-logsin Claude Code to confirm it loaded.

Bundled files

References and scripts installed along with SKILL.md.

references/
RECIPES.md
REFERENCE.md
scripts/
setup.ps1
setup.sh
scripts/lib/
dotnet-runtime.ps1
dotnet-runtime.sh
dotnet-tools.ps1
dotnet-tools.sh

Details

From the project README

SolarWinds Log Search

Search DealerVision production logs through the SolarWinds Observability API using the logs CLI tool.

First-Time Setup (On Demand)

This skill requires the logs CLI tool, .NET 10, and a SolarWinds API token. Install only when the skill is activated.

1. Check if Already Installed

Linux/macOS:

logs --help 2>/dev/null || echo "Not installed - setup required"
echo "$SOLARWINDS_API_TOKEN"

Windows (PowerShell):

logs --help 2>$null
Write-Host "SOLARWINDS_API_TOKEN=$env:SOLARWINDS_API_TOKEN"
2. Install Dependencies and Tool (One-Time Setup)

Run the setup script from the deployed skill folder:

Linux/macOS:

bash ~/.codex/skills/solarwinds-logs/scripts/setup.sh

Windows (PowerShell):

pwsh ~/.codex/skills/solarwinds-logs/scripts/setup.ps1

If you are using Claude Code, replace ~/.codex/skills with ~/.claude/skills.

What this does:

  • Installs .NET SDK 10 if missing (user-level)
  • Installs the logs CLI tool from the skill-local package
  • Verifies the installation
  • Checks for required environment variables

Note: Installation only happens once; the tool becomes globally available.

3. Configure Environment Variable

The tool requires a SolarWinds API token for authentication.

Linux/macOS:

export SOLARWINDS_API_TOKEN="your-token-here"

Windows (PowerShell):

$env:SOLARWINDS_API_TOKEN = "your-token-here"

To obtain a token:

  1. Log in to SolarWinds Observability
  2. Navigate to Settings -> API Tokens
  3. Create a new token with 'Logs Read' permission
4. Verify Setup

Linux/macOS:

dotnet tool list --global | grep SolarWindsLogSearch
logs "test" --limit 1

Windows (PowerShell):

dotnet tool list --global | Select-String SolarWindsLogSearch
logs "test" --limit 1

Prerequisites

  • .NET SDK 10.0+ (required to install the tool)
  • SOLARWINDS_API_TOKEN environment variable (required to use the tool)
  • Default data center: na-01

Alternative: Manual Installation

If you prefer to install manually without the setup script:

# Requires .NET SDK 10.0+
dotnet tool install --global DealerVision.SolarWindsLogSearch --version 2.4.0 --add-source ~/.codex/skills/solarwinds-logs/tools

# If using Claude Code, replace ~/.codex/skills with ~/.claude/skills
# Windows path: %USERPROFILE%\\.codex\\skills\\solarwinds-logs\\tools (or .claude for Claude Code)

# Verify
logs --help

⚠️ Critical: Prefer Simple Searches (No Time Arguments)

The SolarWinds/Papertrail API is finicky with time ranges and often skips recent entries when time arguments are used. The logs tool is designed to start from the most recent entries and page backward through time automatically until a predefined limit (typically 24 hours worth of data).

Default Search Pattern

Always prefer simple keyword searches without time arguments:

# ✅ RECOMMENDED - Simple searches that capture current entries
logs 'error'
logs 'exception'
logs 'timeout'
logs 'DbUpdateException'

# ✅ GOOD - Add filters, but avoid time arguments
logs 'error' --severity ERROR
logs 'exception' --program webhook-api
Why Avoid Time Arguments?

When you use --time-range, --start-time, or --end-time, the API may return a subset of entries that misses vital current/recent log entries. This is a known quirk of the underlying Papertrail API.

# ⚠️ AVOID - Time arguments can miss recent entries
logs "error" --time-range 1h      # May miss errors from the last few minutes
logs "error" --time-range 24h     # May skip entries that just happened
How to Handle User Requests with Time References

When a user asks for something like "errors from today" or "logs from the last hour":

  1. Default to simple search: Run logs 'error' without time arguments
  2. Filter results after retrieval: The agent can filter out older entries from the results
  3. Only use time arguments if the user explicitly wants to exclude current/recent results

Example interpretations:

  • "Search logs for errors from today" → logs 'error' OR 'exception' (filter results yourself)
  • "Find exceptions in the last 4 hours" → logs 'exception' (filter results yourself)
  • "Show me errors, but only from yesterday, not today" → Time arguments are appropriate here
When Time Arguments Are Appropriate

Only use time arguments when the user explicitly wants to:

  • Exclude current/recent results
  • Look at a specific historical window (e.g., "what happened last Tuesday between 2-4pm")
  • Investigate a known past incident with specific timestamps

Quick Commands

# Search for errors (recommended - no time arguments)
logs 'error'

# Find specific exceptions
logs 'DbUpdateException' --severity ERROR --limit 10

# Filter by service (still no time arguments)
logs 'timeout' --program webhook-api

# Get full details for a specific log entry
logs --id 1901790063029837827 --with-data

# Export large result sets to file
logs 'exception' --output-file results.json

Key Options

OptionDescription
--time-range1h, 4h, 12h, 24h, 2d, 7d, 30d
--severityINFO, WARN, ERROR, DEBUG
--programFilter by service name (e.g., media-processing)
--hostnameFilter by host
--limitMax results (default 1000, max 50000)
--with-dataInclude structured JSON payload
--no-dataExclude payload (faster, smaller response)
--output-fileSave full results to file

Workflow Strategy

  1. Start broad - Run initial search without many filters
  2. Narrow progressively - Add severity, time-range, or program filters based on results
  3. Retrieve details - Use --id with --with-data for full log entry inspection
  4. Export large datasets - Use --output-file for comprehensive analysis

Output Format

Returns JSON with:

  • success: boolean status
  • query: search parameters used
  • summary: statistics including severity breakdown, truncation info
  • results: array of log entries with timestamps, source, severity, message
  • pagination: info for getting more results

Advanced Usage

For detailed documentation on query syntax, MCP server modes, and advanced patterns, see references/REFERENCE.md.

For common investigation patterns and recipes, see references/RECIPES.md.

Comments

Markdown supported