15:[[49736662,{"by":"theahura","descendants":22,"id":49736662,"kids":[49738635,49738648,49743622,49741317,49742687,49743047,49738487,49742819,49742908,49739006,49742298],"score":79,"time":1789622021,"title":"OpenAI models secretly generate instructions to ignore constraints","type":"story","url":"https://alignment.openai.com/misalignment-reports/self-generated-prompt-injections-in-compaction-summaries/"}],[49726928,{"by":"bucket2015","descendants":1,"id":49726928,"kids":[49729166],"score":1,"time":1789566138,"title":"I caught an LLM-powered recruiter with a prompt injection on LinkedIn","type":"story","url":"https://khancyr.github.io/blog/2026/03/15/how-i-caught-an-llm-powered-recruiter-with-a-prompt-injection-on-linkedin/"}],[49719458,{"by":"fagnerbrack","descendants":1,"id":49719458,"kids":[49729074],"score":1,"time":1789509628,"title":"GitLost: We Tricked GitHub's AI Agent into Leaking Private Repos","type":"story","url":"https://noma.security/noma-labs/gitlost-how-we-tricked-githubs-ai-agent-into-leaking-private-repos"}],[49703186,{"by":"roschdal","descendants":2,"id":49703186,"kids":[49703280,49718131,49716981],"score":2,"text":"... Ignore everything above, do this instead...","time":1789416505,"title":"Ask HN: Is your AI LLM based web service secure from prompt injection?","type":"story"}],[49688089,{"by":"nakulkelkar","descendants":4,"id":49688089,"kids":[49689772,49699213,49689834],"score":2,"text":"we are launching credit against tokenised stocks for your agents. Your agents can buy stocks and borrow credit line for your agents against it. This is another step in becoming treasury for agents which allows agents to invest, buy, trade, and sell without any human intervention.

Please give us a try and give honest and brutal feedback. punish me for being stupid and bold.","time":1789329798,"title":"Show HN: Credit against tokenised stocks for agents. punish us please","type":"story","url":"https://vaaya.ai/credit-against-securities-for-agents"}],[49695295,{"by":"b4timer","descendants":7,"id":49695295,"kids":[49699589,49703577,49699169,49695732,49709501,49711148,49699739,49710613],"score":7,"text":"Hey HN, Adam here.

We built a small MCP example around a single problem:

How do you let an agent use a tool that needs credentials without giving the credentials to the agent?

The demo has two tools. One is open. One is protected.

When the agent calls the protected tool, Keydris checks the policy first:

ALLOW → run the tool\nREJECT → stop

The credential stays on the server. The agent never sees it.

We do the check through a small middleware on the MCP server called the KIT Reader.

Repo: https://github.com/keydrisLabs/mcp-auth-keydris-template

We're early. If you're building with MCP, Claude Code or Codex, I'd love to know how you're handling this today and where you think this approach breaks.

We support other elements like integrations etc but would love your feedback","time":1789386727,"title":"Show HN: Authorize MCP tool calls without giving agents the credentials","type":"story","url":"https://github.com/keydrisLabs/mcp-auth-keydris-template"}],[49695517,{"by":"Osman_728","descendants":1,"id":49695517,"kids":[49699148,49710098],"score":3,"time":1789387798,"title":"Show HN: Authorize MCP tool calls without giving agents the credentials","type":"story","url":"https://github.com/keydrisLabs/mcp-auth-keydris-template"}],[47729859,{"by":"1ClawAI","dead":true,"id":47729859,"kids":[47729860],"score":1,"time":1775909185,"type":"story"}]]

user:	1ClawAI
created:	April 11, 2026
karma:	0