AI Aulis

AI NEWS IN ENGLISH · 19 November 2026

The most important AI news every morning. AI writes and checks the stories against the original sources – check important facts in the sources.

38 stories · next issue tomorrow morning · How the paper is made →

← Issue 19 November 2026Paper

LangSmith links coding agent costs to the pull requests they open

AI-generated contentWritten by AI

youtube.com, docs.langchain.com · 2 sources · Sources published: 18 November 2026 · Some source publication dates are unknown

A new beta feature in LangChain’s LangSmith links coding agents’ costs to GitHub and GitLab pull requests. It shows what a merged change cost and how much of the spending did not lead to a pull request.

In a video published on 18 November, LangChain presented a new view in LangSmith. The view brings the costs of Claude Code, OpenAI Codex, Cursor, GitHub Copilot and Deep Agents into one place, so teams do not have to look them up in each vendor’s own admin console. Coding is the largest category of enterprise AI application spend, at $34 billion in 2026, the company said. Yet most engineering leaders cannot say exactly what the money buys, it added.

The feature relies on traces, which record an agent’s work steps. GitHub or GitLab notifies LangSmith when a pull request is opened, updated or merged. LangSmith then links that pull request to the threads whose traces record the same repository and source branch. This shows what a merged pull request cost and how much of the spending never led to any pull request.

The link is therefore only as complete as the Git data the agents record. LangChain’s documentation says GitHub Copilot records the repository and branch only in the Local harness, and even then only in a folder with a Git remote. A harness is the program that controls an agent: it gives the model its tools and runs its work steps. The view does not show pull requests merged before the connection was set up, so per-pull-request costs only start to accumulate after that.

Once GitHub or GitLab is connected, the overview shows cost per merged pull request, cache share, time to merge and merge rate, among other figures. Charts can be grouped by user, repository, harness or model, for example. If a team switches models, the documentation advises comparing costs before and after the switch. It also recommends checking whether pull requests are still being merged over the same period.

The feature is in public beta on LangSmith Cloud. It requires an agent-based workspace and is not available on self-hosted LangSmith. The pull request connection and coding charts appear only on agents whose type is set to Coding. Every coding agent the team uses needs tracing set up to send its data to such an agent. For Claude Code, Codex and Cursor, this means installing a LangSmith plugin. Changing the type of an existing agent requires permission to manage workspace settings. Connecting a GitHub or GitLab account requires permission to update all of the agent’s environments, and installing the GitHub app may need approval from an organisation administrator.

Report an error in this story