Pike Agent

pike-agent records and analyzes how programs behave on Linux. It traces a program's activity, indexes it into a database, and lets you chat with an LLM agent about it in a TUI.

Example of prompts:

• Crash diagnosis: This program crashed with a bus error. What happened?
• Race condition detection: Are there any race conditions in how this program creates temporary files?
• Malware triage: I found this process running on a production server. Should I be worried?

See Say hi to Pike! for an introduction.

How it works

pike-agent uses strace to record every system call a program makes, then indexes them into a SQLite database with full-text search. An LLM agent queries this database through tool calls to answer your questions about the program's behavior.

• Supports local inference (llama.cpp) and external APIs (Claude, Gemini, etc.)
• Full-text search over syscall arguments
• Man page lookup for syscall documentation

Installation

uv pip install -e .

Requires Python 3.13+ and strace installed on the system.

Usage

On first run, a default configuration file is created at ~/.config/pike-agent/config.toml. Edit it to configure your inference provider and model. By default it points to a local llama.cpp server.

Pike-agent works best with models that are good at tool calling, for example: Gemini 3 Pro, Qwen 3.5, Claude 4.6 Sonnet...

Recording a trace

Trace a command:

pike-agent run -- ./my-program --flag

Attach to a running process:

pike-agent attach 1234

Both produce a .db file with the indexed trace.

Chatting with the agent

Start a chat session on a recorded trace:

pike-agent chat trace.db

License

GPLv3.