Aldena is an AI coding agent orchestration layer that takes work from an idea in Jira or Linear to a ready-to-merge pull request. For founder Aaron Delasy, dependable access to the live web is a core part of helping those agents find relevant information and finish the work they are given.
The challenge was consistency. Some models include their own web tools, while others do not. Aldena needed one shared search and fetch layer that could work across every model on the platform.
"Some of the models don't provide these tools out of the box, so we created our own tools instead to make sure every LLM has a unified mechanism to access the web."
One web interface for every model
Aldena exposes web_search and web_fetch tools to its agents. That gives each model the same route to current search results and page content, regardless of the web capabilities included by the model provider.
The team already had another provider behind those tools. Context.dev could take its place without forcing Aldena to redesign the agent interface or maintain a different integration for each model.
"Context.dev allowed me to enable
web_fetchandweb_searchtools for my platform."
A provider replacement completed in 30 minutes
Aaron delegated the migration to Claude and pointed it at the Context.dev documentation. The agent replaced the existing provider, then completed the normal engineering checks around the change in the same session.
"It was super easy. I just pointed Claude at the documentation, and it figured out everything in about 30 minutes, including linting, type checking, unit tests, and end-to-end tests. We already had another provider, so we just replaced it."
The only difference was a published date supplied by the previous provider. Aldena had used it internally to judge freshness, but the value was not shown to customers. Rather than hold up the migration for an unused field, the team removed that part of the implementation.
"The only thing that agent wasn't able to figure out is that our previous provider was providing published dates, which we used to understand how stale the data is. We just removed that part because we weren't showing it to the end consumer anyway."
Live web search inside an agent workflow
With Context.dev connected, an Aldena agent can search the web as part of a larger task and continue working from the results. In the example below, an Airbnb Finder agent runs multiple searches for an affordable San Francisco stay before turning the results into recommendations.
The web step stays inside the agent experience. The model decides what to search for, Context.dev retrieves the live results, and the agent uses that context to produce the next part of its response.
Current context from idea to pull request
Aldena is designed to move work from an issue in Jira or Linear to a pull request that is ready for review. Agents handling that process need more than the information already inside a model. They also need a dependable way to retrieve relevant material from the web while they work.
"Aldena is an AI coding agents orchestration layer that allows you to go from an idea in Jira or Linear to a ready-to-merge PR. It's crucial to give agents access to the internet and be able to fetch relevant information."
By standardizing that access through Context.dev, Aldena can give every supported model the same search and fetch tools without treating the live web as a model-specific feature.
Key benefits
- 30 minutes from docs to tested integration: Claude replaced the previous provider and completed linting, type checks, unit tests, and end-to-end tests in one session.
- One tool surface across models: every agent can use the same
web_searchandweb_fetchinterface, even when the underlying model does not include web access. - A direct provider replacement: Aldena switched the service behind its existing tools without redesigning the agent workflow.
- Live context inside larger tasks: agents can search for current information, fetch relevant pages, and continue working from what they find.
The outcome
Aldena gave its AI coding agents a consistent route to the live web without creating a separate implementation for every model. The migration took about 30 minutes, included the team's normal quality checks, and left the platform with unified web search and fetch tools built on Context.dev.
Building agents that need the same web access across models? Context.dev's Web Search API finds current sources, while the Markdown API turns the pages behind them into clean content for your agents.
P.S. Want to move work from an idea to a ready-to-review pull request? Explore Aldena.