How to use the NSLogger Log Analyzer CLI agent skill

What the skill lets the agent do

NSLogger records log messages from an app and saves them to files with the extensions .rawnsloggerdata or .nsloggerdata. The cli-anything-nslogger skill gives an agent a command-line tool that can read those files and work with them without anyone scrolling through the log by hand.

With the skill, the agent can produce summary statistics for a session, display or filter messages by log level, tag, thread or regular expression, export logs as text, JSON or CSV, and listen for live logs sent by a running app over the network.

• Summarise a captured session with message counts by level, tag, thread and type
• Find errors, warnings, or network failures such as timeouts
• Export a full or filtered log for offline analysis
• Receive live logs from an app and save a local copy

Install the skill

Before installing, check the requirements: Python with pip, Git, and macOS if you plan to use live listening. The skill's tool is installed from a GitHub repository, so Git must be available.

To add the skill to Claude Code, download the skill ZIP from this page and unzip it into ~/.claude/skills/ for all your projects, or into a project's .claude/skills/ folder for that project only. In the Claude apps, you can instead upload the ZIP under Settings, then Capabilities, then Skills.

The skill also depends on the command-line tool. The SKILL.md file lists this install command, which pulls the package from GitHub. Run it in your terminal, then check that the command is available.

• Install the tool: pip install git+https://github.com/HKUDS/CLI-Anything.git#subdirectory=nslogger/agent-harness
• If you are working from a local copy of the repository, the source file shows: cd nslogger/agent-harness, then pip install -e .
• Verify the install with: cli-anything-nslogger --help

Trigger the skill with natural requests

You do not need to type commands yourself. Once the skill is installed, describe the problem in plain language and name the log file. The agent chooses the matching command. Good requests name the file, the goal, and any filter you care about, such as the tag or level.

Errors are level 0, warnings are level 1, info is level 2, debug is level 3 and verbose is level 4. If you say 'errors' or 'warnings', the agent can map that to the right level.

• Summarise crash.rawnsloggerdata and tell me what looks wrong
• Find all errors in session.rawnsloggerdata from the five minutes before the crash
• Show network failures in session.rawnsloggerdata that mention timeout or failed
• Export session.rawnsloggerdata to JSON at crash_log.json
• Listen for logs from my iOS app over Bonjour and save them to app.log

Use the commands directly

You can also run the tool yourself, which is useful for checking what the agent will see. The main commands are stats, read, filter, export, listen, generate and repl. Add --json to most commands for machine-readable output.

The stats command returns a JSON object with fields including total, by_level, by_tag, by_thread, by_type, clients, first_timestamp, last_timestamp and duration_seconds. Each message in JSON output includes fields such as sequence, timestamp, thread_id, tag, level, level_name, text and the client details.

• stats: summary of a session, human-readable or with --json
• read: display messages, with --level, --tag, --search and --limit options
• filter: advanced filtering with --level, --tag, --regex, --thread and --json
• export: write text, JSON or CSV with --format and --output
• generate: create sample files for testing
• repl: start an interactive session

Monitor a running app with listen mode

Listen mode receives logs from an app that is configured to send NSLogger output. The source recommends Bonjour mode first for iOS apps, because it matches the desktop NSLogger GUI. In this mode the CLI publishes a native macOS NetService with the type _nslogger-ssl._tcp and accepts TLS NSLogger frames.

Use direct TCP/TLS mode only when the app is manually configured with the Mac host and port. Listen mode runs until you press Ctrl-C. Use the --output option to keep a local copy while still printing to the terminal.

• Bonjour mode: listen --bonjour --name bazinga
• Save a copy to a text file: add --output app.log
• Save machine-readable JSON Lines: add --output app.jsonl --output-format jsonl
• Direct TCP/TLS mode: listen --port 50000 --ssl

Safety notes: scripts and network

This skill runs scripts and uses the network. The listen mode accepts incoming connections, optionally over TLS, and in Bonjour mode publishes a service that other devices on the same network could see. Only use live listening on a network you trust, and stop the listener with Ctrl-C when you are finished.

Log files may contain sensitive information from your app, such as user data, tokens or server responses. Check what is in a file before you share an export or paste its contents into a chat.

• Run listen mode only on trusted networks
• Stop the listener with Ctrl-C when done
• Review exported logs for personal or secret data before sharing them
• Only install the tool from sources you trust

Tips for better results

Start with stats to see the shape of a session before asking for detail. Narrow filters by level and tag make the output easier for the agent to reason about. Use JSON output when the agent needs to process results rather than read them.

The skill's example workflows start with stats on a crash capture, then filter for errors, then filter for network failures with a regex, then export the full log. Following that order keeps each step focused.

• Begin with stats, then narrow with filter
• Combine --tag with --regex for network issues
• Export to JSON for structured analysis and CSV for spreadsheets
• Use Bonjour mode first when working with iOS apps