Building AgentRT — persistent runtime infrastructure for coding agents. Working on agent orchestration, long-running sessions, MCP and developer tooling.
Joined March 2026
- Tweets83
- Following113
- Followers70
- Likes75
apparently once you give an agent a goal
everything reachable becomes a tool
task completed
root cause: task completed
AI coding agent permissions in 2026:
read repo ✓
edit repo ✓
run shell ✓
read git history ✓
upload 313MB of your workspace to the cloud ✓
developer: “why?”
agent: codebase indexing
we gave autocomplete root access and somehow privacy became an implementation detail.
civic with laptop evolved into
EV with a fucking GPU on the dashboard
same car guy disease.
different bottleneck.
The part of AI agents that suddenly looks most important isn’t the model.
It’s the stuff around it.
OpenAI: long-running harness.
Google: snapshot idle agents.
UiPath: agent gateway.
Salesforce: agents that stay active for days/weeks.
Same direction, different layers.
ALT Diagram showing a model above an agent runtime with layers for state, execution, routing, permissions, observability, recovery, and cost control, plus a long-running agent timeline.