All news briefs
Product update

Why AI coding harnesses disagree over the right code context

The story in five sentences

  1. Ars Technica compares two approaches to supplying context for AI coding agents.
  2. Claude Code usually searches for the code it needs while working on a task.
  3. Augment Code indexes repositories in advance with embeddings, a retrieval model, and a vector database.
  4. Augment reports similar accuracy with 33 percent fewer tokens in its own Terminal-Bench comparison.
  5. The article warns that more autonomous agents do not replace human engineering judgment.

Full explanation

Ars Technica contrasts two philosophies for the harness surrounding an AI coding agent. The harness determines which code the model can see, which tools it can use, and how it interacts with a project. Anthropic keeps Claude Code comparatively lean and lets the agent search for relevant files as a task unfolds. Augment Code instead prepares repository context in advance with embeddings, a retrieval model, and a vector database.

Augment vice president Vinay Perneti argues that this indexed approach is especially useful for large private repositories. In the company’s own Terminal-Bench comparison, Augment reported a similar success rate while using 33 percent fewer tokens. That is a vendor benchmark rather than an independent evaluation. The compared setups also used partly different tools and optimization targets, which limits direct conclusions about relative quality.

Broader autonomy does not remove the need for experienced engineers. Humans still have to define useful specifications and assess whether an implementation fits the wider system. Coding agents may duplicate existing functionality when they lack architectural understanding. The expectation that inexpensive open-weight models will eventually handle more routine work is presented as a plausible direction, not a confirmed outcome.

Privacy setting

May we measure pseudonymous page views and session replays?

With your consent, PostHog EU measures which pages are opened and additionally records your session: mouse movement, clicks, scrolling and the rendered page content are stored as a replayable reconstruction. Typed input is masked before sending. The contents of sign-in, sign-up, password recovery, the founder chat, the newsletter field, your email display, your quiz answers and the checklists are not recorded at all; on the sign-in and sign-up pages recording is paused. Addresses are always stored without query parameters. A random device identifier and technical connection data such as the IP address are processed as well. In addition, strictly defined usage events are measured: how far a post was read, which article element was used, which call to action was clicked, how a newsletter sign-up ended, how far you get in a course or in the feed (recording only whether a task was correct or incorrect, never your answer itself), which video you start, whether you switch the language, and how a sign-in or sign-up attempt ended — without the email address, without the password, and without the error message. Only values from a fixed list and whole numbers from fixed ranges are transmitted — no input and no free text. We also label production, internal, and automated test visits separately and derive whether a visit came from a known interface such as ChatGPT, Claude, or Perplexity from a known referring domain or a strictly allowed campaign value. The full referring address and query parameters are not sent, and this cannot identify a specific AI model. If PostHog's IP-based geo enrichment is enabled in the project, the service can derive an approximate country, continent, and region; we do not request browser or GPS location. Every visit is also assigned to a page area from a fixed list (for example home, blog, tools, account) — the area is sent, not the address. Page loading and stability metrics are measured as well (Core Web Vitals: LCP, CLS, INP, FCP), without network payloads. A click on a link leading away from the site is recorded only as the target domain from a fixed list, without path, query parameters, or link text. For the Idea Wheel, AI Labelling and Limit Reset tools only the kind of action is measured — spin, option changed, copy or export, for instance — never your input and never a result. Learn more about privacy

Without your consent no analytics code is loaded and nothing is recorded. You can withdraw at any time: withdrawal ends collection immediately; data already collected may have been transmitted by then and is deleted after the storage period.