02 of 04 · Outside
External Context Feeds
External Sources
Agents need continuous external awareness. We create a discrete feed for each external source you care about. One build is technical intelligence, indexing open-source software and research, scored for agents to evaluate directly. Our first build scored over 100,000 entities and gave agents structured data on what is defensible, what is moving, and what is risky.
- Builds a discrete feed per external source
- Scores new implementations in real time
- Feeds agents live, structured intelligence
- Enables direct code and pattern reuse
Agents need continuous awareness. A model trained last year does not know what shipped this week.
We create a discrete feed for each external source you care about. Each one is structured, scored, and live, and each one is served to your agents the way they read any other tool.
Technical intelligence
A technical intelligence feed indexes open-source science and engineering repositories, scoring new implementations as they appear so agents can evaluate them directly. Our first build scored over 100,000 entities. Each score answers three questions.
- What is defensible.
- What is moving.
- What is risky.
This feed helps agents reuse proven open-source implementations instead of writing redundant code.
Any external source
Other external sources, such as OSV vulnerability data, become their own feeds. One source, one feed, one shape.
Where it sits
This is the context of the outside. Internal context feeds is the context of the inside. Context management sits between them.
Questions.
What is an external context feed?
A discrete, structured feed we create for one external source, so agents can query real-world changes outside your walls.
What kind of feed can you build?
Technical intelligence is one example. It monitors open-source repositories and research to score new implementations for agents. Our first build scored over 100,000 entities on defensibility, momentum, and risk.
Which other sources can become feeds?
Other external sources, such as OSV vulnerability data, become their own feeds. We create a discrete feed for each.
Can agents call a feed directly?
Yes. Feeds are served over MCP and API, so agents query them directly. Our first technical intelligence feed tracked 525 technical domains.
Next module · Between
Context Management