solarwinds-logs

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:

```bash

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

echo "$SOLARWINDS_API_TOKEN"

```

Windows (PowerShell):

```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

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

```

Windows (PowerShell):

```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:

```bash

export SOLARWINDS_API_TOKEN="your-token-here"

```

Windows (PowerShell):

```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:

```bash

dotnet tool list --global | grep SolarWindsLogSearch

logs "test" --limit 1

```

Windows (PowerShell):

```powershell

dotnet tool list --global | Select-String SolarWindsLogSearch

logs "test" --limit 1

```

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

If you prefer to install manually without the setup script:

```bash

# 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

```

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:

```bash

# ✅ 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.

```bash

# ⚠️ 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

```bash

# 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

```

| Option | Description |

|--------|-------------|

| --time-range | 1h, 4h, 12h, 24h, 2d, 7d, 30d |

| --severity | INFO, WARN, ERROR, DEBUG |

| --program | Filter by service name (e.g., media-processing) |

| --hostname | Filter by host |

| --limit | Max results (default 1000, max 50000) |

| --with-data | Include structured JSON payload |

| --no-data | Exclude payload (faster, smaller response) |

| --output-file | Save full results to file |

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

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

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

For common investigation patterns and recipes, see [references/RECIPES.md](references/RECIPES.md).