Brian Hung
Projects
Harness
Recursive language models and a continual harness, as MCP servers any agent can use.
rlm_deep_query
the context never enters the model window
question
answer
rlm_store_context
full text
code
stdout, first 4,000 chars
slice
answer
MCP host
Claude Code, Codex
context store
rlm_store_context
root model
generateText tool loop
Python kernel
PROMPT holds the context
sub-call
one depth down
The answer is the model's reply, or FINAL from the kernel when it is too long to write out.
Below the depth limit a sub-call gets its own kernel; at the limit it is one plain model call.
On Docker and Cloudflare, the model calls a recurse tool instead.