rastriga

7 ms·

1. 1.▲
   by rastriga 8mo ago

   Part 2 of a small personal investigation into how Claude Code actually works. This one focuses on the system prompt and how much behavioral governance is encoded there. Part 3 will look at tool execution.
2. 2.▲
   Claude Code’s system prompt (verbatim) and how it shapes the agent(rastrigin.systems)

   1 points by rastriga 8mo ago|1 comments
3. 3.▲
   by rastriga 8mo ago

   Fair point — the stateless prompt+context pattern itself is standard, and the headline probably leans too much into “surprise.” The takeaway for me was the opposite of hidden complexity: start simple, don’t over-engineer. Claude Code feels
4. 4.▲
   by rastriga 8mo ago

   Thanks — glad it resonated! Part 2 should uncover a lot of the magic behind the scenes. And thanks for sharing the link. Running claude code against a local llm is a really interesting direction, but I need more RAM...
5. 5.▲
   by rastriga 8mo ago

   Thanks! I tried doing a similar comparison with Codex CLI and Cursor, but hit a wall. Codex doesn’t seem to respect standard proxy env vars, and Cursor uses gRPC. Claude Code was the only one that was straightforward to inspect. Opencode lo
6. 6.▲
   by rastriga 8mo ago

   You can change the system prompt claude code sends, which changes how the agent frames behavior, but claude still has internal and server side safety layers. So removing or rewriting the client system prompt won't allow to magically by
7. 7.▲
   by rastriga 8mo ago

   Statelessness simplifies scaling and operational complexity. They cache the system prompt, but otherwise each request is fully self-contained. It’s an obvious tradeoff, and I wouldn’t be surprised if they move toward some form of server-sid
8. 8.▲
   by rastriga 8mo ago

   It's the last 5 commits, not the full history. Here's what actually gets sent in the system prompt: gitStatus: This is the git status at the start of the conversation... Current branch: main Main branch: main Status: (cl
9. 9.▲
   by rastriga 8mo ago

   thank you!
10. 10.▲
    by rastriga 8mo ago

    yes, it's obvious the data goes to Anthropic. What wasn't obvious to me was what exactly is included and how it's structured: system prompt size, full conversation replay, file contents, git history, tool calls. The goal was
11. 11.▲
    by rastriga 8mo ago

    I built a MITM proxy to inspect Claude Code’s network traffic and was surprised by how much context is sent on every request. This is Part 1 of a 4-part series focusing only on the wire format and transport layer. Happy to answer technical
12. 12.▲
    What Claude Code Sends to the Cloud(rastrigin.systems)

    42 points by rastriga 8mo ago|21 comments