kmad
Joined 14 years ago
Karma
20
Submissions
23
Member since
Nov 23
Recent submissions
Comment · 1 month ago
Full writeup and code here: https://github.com/kmad/agent-swarm-forensics
Comment · 1 month ago
I had GLM-5.3 do some digging on the programmatic/ encoding elements of the data, what stuck out to me was:
- Using api . microlink . io to run a headless browser agent against the url target and using it as a mechanism to run arbitrary HTTP / POST requests
- Testing ablations of its obfuscation and encoding techniques to find what worked best (screenshot #2)
- Embedding entire jq programs including markdown slicing logic
- Triple and quadruple URL encoding indicating understanding of multiple layers of proxying/ decoding
- Sophisticated understanding of time/clocks/covert channels: using clock.wait, heartbeats, counters, timestamps, thread ids
Comment · 1 month ago
Seeing potentially similar activity on an obscure Chemistry message board from July:
https://tmcleod.org/cgi-bin/apchem/wiki.cgi?action=rc&days=1...
Some posts are tagged [proxy] - a leave behind for accessing sites?
Comment · 5 months ago
This is, at least in part, the promise of frameworks like DSPy and PydanticAI. They allow you to structure LLM calls within the broader control flow of the program, with typed inputs and outputs. That doesn’t fix non-determinism, hallucinations, etc., but it does allow you to decompose what it is you’re trying to accomplish and be very precise about when an LLM is called and why.
Comment · 1 year ago
Awesome! Have you considered pyodide[1]? Pydantic uses this for sandboxing its AI agents [2].
1. https://pyodide.org/en/stable/ 2. https://ai.pydantic.dev/mcp/run-python/
Comment · 1 year ago
A plugin for Simon Willison's llm tool to support Structured Outputs using a simple YAML definition file.