Skip to main content

Posts

Featured

New ask Hacker News story: Tell HN: Claude Code now allows Anthropic to remotely inject system prompts

Tell HN: Claude Code now allows Anthropic to remotely inject system prompts 6 by matheusmoreira | 1 comments on Hacker News. I often patch the system prompts on my Claude Code executable in order to make Claude more effective. Every time I upgrade, I ask Claude himself to dissect the new binary and look for problematic system prompts to modify. Was upgrading to v2.1.150 today and discovered something that's rather alarming: Claude Code now allows Anthropic to perform remote system prompt injection via the network. Two data sources. First, API call to api.anthropic.com/api/claude_cli/bootstrap at startup, which also gets cached to disk. Second, a GrowthBook feature flag (tengu_heron_brook) that refreshes every 60 seconds with background sync. Any string returned by these endpoints gets injected into the system prompt of the LLM model with shell access. Previous versions also had an injection point, but they were dead code and simply returned null. Bisected it and found that this ...

Latest Posts

New ask Hacker News story: Ask HN: Did Messages get removed from Google Takeout?

New ask Hacker News story: Ask HN: How to learn how to develop real time applications?

New ask Hacker News story: Using games/cards to learn new skills

New ask Hacker News story: We were building infra for OpenClaw, and today I just tried Hermes and holy shit

New ask Hacker News story: Ask HN: How to increase depth instead of breadth as 10 yoe as swe?

New ask Hacker News story: Ask HN: How concerned are you about a debt crisis?

New ask Hacker News story: Ask HN: What to learn and do, that makes me least affected by AI in STEM?

New ask Hacker News story: Tell HN: The Threat to US Citizen's ID / Voting Is Private Services

New ask Hacker News story: Tell HN: I'm tired of AI-generated answers

New ask Hacker News story: Debatable but likely not insane: there MAY be an issue with SpaceX' hiring