Claude Code vs. Augment Code: The Debate Over AI Coding Harness Design

Anthropic bets on a lean, grep-based harness for Claude Code; Augment Code argues semantic retrieval delivers better results in large private codebases.

The most consequential design decisions in AI-assisted software development may now live not in the underlying models, but in the harness software wrapped around them — and two companies are making sharply different bets on how that layer should work [1].

A harness is the software that determines what an AI model sees, what actions it can take, and how it interacts with a codebase [1]. As Ars Technica reported on July 20, the debate has sharpened between Anthropic’s Claude Code, which favors a minimal harness, and Augment Code, which pre-indexes repositories using semantic retrieval before an agent ever touches a task [1].

The lean harness argument

Cat Wu, head of product for Claude Code at Anthropic, told Ars Technica that the product team deliberately avoids building structured context around a codebase in advance by default [1]. Her reasoning centers on model velocity: because underlying models are improving so quickly, locking in opinionated features risks being overtaken by the models themselves [1].

On the question of whether semantic code-navigation tools produce measurable gains, Wu was direct. “Going by the evals, we don’t see a measurable change,” she said, adding that the team prefers “shipping a leaner harness with fewer opinionated tools and just letting developers add their own if they want” [1].

Claude Code’s current approach relies on grep-based context retrieval — pattern-matching searches through code — rather than semantic or vector-based methods [1]. OpenAI’s Codex takes a similar approach, according to Ars Technica’s reporting [1].

The semantic retrieval counter-argument

Augment Code, founded in 2022, spent roughly 18 months before the public release of ChatGPT researching retrieval and embedding models specifically for large codebases, according to Vinay Perneti, the company’s VP of Engineering [1]. The system pre-indexes a repository using an embedding and retrieval model pair, a vector database, and a back-end optimized to return results in sub-milliseconds [1].

Perneti argues the advantage is most pronounced in large, private codebases — precisely because public, open-source repositories are often already memorized by frontier models through training data [1]. “When you’re doing this in a private repo, a model has never seen that repo,” he told Ars Technica, explaining that semantic pre-indexing shortens the iteration loop for finding relevant code [1].

On token efficiency, Perneti pointed to a benchmark the company published: running Terminal-Bench with the same underlying model, Augment Code completed tasks at similar accuracy to Claude Code but used 33 percent fewer tokens [1]. He attributed the savings to spending less of the token budget on exploratory context-gathering [1].

Are they measuring the same thing?

When Ars Technica put Anthropic’s eval results directly to Perneti — that adding semantic retrieval tools showed no measurable improvement — he questioned whether the implementations being compared were equivalent [1]. “Not all retrieval systems are equal just like not all databases are equal,” he said, suggesting that a generic retrieval-augmented generation (RAG) implementation layered onto Claude Code would differ substantially from Augment’s purpose-built context engine [1].

Perneti also pushed back on the idea that rapidly improving model intelligence makes harness design irrelevant. His framing separates two ingredients required for quality outcomes: intelligence, which scales with model improvements, and context, which does not automatically improve just because a model gets smarter [1]. “How much of your token budget is going towards context gathering and producing the right outcomes?” he said, describing harness design as a systems engineering problem of balancing those two dimensions at minimum cost [1].

Beyond Claude Code and Augment Code, the harness space includes Google’s Antigravity, the open-source OpenCode, and Cursor, each with different design philosophies around how models are deployed in agentic workflows [1].


Sources

  1. Ars Technica — Beyond grep: The case for a context-rich AI coding harness

This article was drafted with AI from the cited sources and checked against them before publication. Spot an error? Let us know.

Anthropic's coding agent can now open, read, click, and type on external websites directly inside the app, with safety classifiers screening write actions.

Anthropic has added an integrated browser to Claude Code, letting the AI agent open, read, click, and type on live external websites without leaving the app [1].

For developers, the practical upside is direct access to documentation sites, issue trackers, and similar resources during a coding session — no more copy-pasting content between a separate browser and the agent [1].

How it works

The browser behaves like a standard tab-based browser and opens via a keyboard shortcut, according to The Decoder’s reporting [1]. Under the hood, Claude uses the same tooling it already relies on to preview local apps, extended with additional safety checks for external sites [1].

Classifiers screen any write actions — such as form submissions — before they execute [1]. Anthropic said Claude will not purchase anything, create accounts, or bypass CAPTCHAs (the challenge-response tests used to block automated access) without explicit user consent [1]. The browser also runs on a clean profile with no saved login credentials, limiting the blast radius if something goes wrong [1].

Controls for organizations

Anthropic announced on X that organizations can restrict the browser to an allowlist of approved external sites, or disable the browser tools entirely [1]. Developers who need Claude to act within their own authenticated sessions — where saved logins matter — are directed to use the Chrome extension instead [1].


Sources

  1. The Decoder — Claude Code now has a built-in browser that lets the AI read, click, and type on external websites

This article was drafted with AI from the cited sources and checked against them before publication. Spot an error? Let us know.

Paid subscribers on Pro, Max, Team, and premium Enterprise seats can use Anthropic's newest model at no extra cost through July 12, 2026, up to half their weekly usage limit.

Anthropic has extended free promotional access to Claude Fable 5 — its newest model — through July 12, 2026 at 11:59 PM Pacific Time, giving paid subscribers more time to test the model without additional charges [1].

The promotion, which originally launched July 1, matters to developers and teams because it lowers the barrier to evaluating a new frontier model mid-cycle, without requiring a separate billing commitment or plan upgrade [1].

Who qualifies and how to access it

Promotional access is available on Pro, Max, Team, and premium seats on seat-based Enterprise plans [1]. Free-plan users are excluded [1]. There is nothing to activate — eligible subscribers can simply select “Fable 5” from the model picker in Claude on the web, Claude Desktop, or Claude Mobile [1].

Developers using Claude Code need version 2.1.170 or later to access Fable 5 [1]. Users of Claude Cowork, Anthropic’s collaborative workspace feature built into Claude Desktop, must be on the latest version of that app [1].

Usage limits and what happens after the promotion

During the promotional period, Fable 5 draws from each subscriber’s existing weekly plan limits, not a separate bucket [1]. Subscribers can use up to 50% of their weekly limit on Fable 5 at no extra cost [1]. Importantly, usage of other Claude models counts against the same overall weekly cap, so the 50% Fable 5 allowance is a ceiling, not an addition [1].

Once a user hits that Fable 5 sub-limit, they have two options: purchase usage credits — billed separately from the subscription — to keep using Fable 5, or switch to another model and continue within their remaining plan limits [1]. If usage credits are already enabled and a user keeps prompting Fable 5 after hitting the limit, billing switches to credits automatically [1].

After July 12, Fable 5 is no longer included in any plan’s weekly limits, and continued access requires usage credits across all plan types [1].

Enterprise seat distinctions

The promotion treats Enterprise seat types differently. Premium seats include Fable 5 access through July 12 at no extra cost, drawing from each member’s seat usage [1]. Standard seats on legacy seat-based Enterprise plans do not include promotional access; all Fable 5 usage on those seats draws from usage credits, meaning the model is unavailable entirely if the organization has not enabled usage credits [1].

API access to Fable 5 is billed separately at standard rates regardless of plan type and is not covered by the promotion [1].

In Claude Code, administrators can restrict Fable 5 availability through managed settings, so users who do not see the model in that environment may have had access blocked by their organization [1].


Sources

  1. support.claude.com — Claude Fable 5 promotional access | Claude Help Center

This article was drafted with AI from the cited sources and checked against them before publication. Spot an error? Let us know.

Built on the MIT-licensed GLM-5.2 model, ZCode offers a full agentic coding workflow at a fraction of the price of Western competitors.

Chinese AI company Z.ai has released ZCode, an agentic coding tool designed to compete directly with Anthropic’s Claude Code and OpenAI’s Codex, priced well below those Western alternatives, according to reporting by The Decoder [1].

The launch matters for developers and engineering teams evaluating AI coding tools: ZCode is built on GLM-5.2, a model that a Snowflake comparison across 103 tasks found nearly tied with Claude Opus 4.7 — at a fraction of the cost [1].

What ZCode Does

ZCode operates as a dedicated agent that handles file access, terminal output, browser context, and Git changes inside a single workflow, the company said [1]. Developers can write, debug, test, and review code using natural language commands [1]. Z.ai says a 1 million-token context window keeps even multi-step programming tasks coherent without losing context [1].

The agent can also be controlled remotely through Feishu, WeChat, or a smartphone, giving teams flexibility in how they integrate it into existing workflows [1].

Pricing and Availability

New customers receive a free five-day trial with up to 5 million tokens per day [1]. Subscribers get roughly 1.5 times more quota through July 2026, according to Z.ai [1].

The Model Behind It

GLM-5.2, the foundation model powering ZCode, shipped in June 2026 under an MIT license — meaning developers can use and modify it freely [1]. It has gained traction among developers who view it as competitive with pricier Western models, and the Snowflake head-to-head across 103 tasks showed GLM-5.2 and Opus 4.7 nearly tied after three attempts [1].

Z.ai had already been competing with Western providers on price through GLM-5.2 before the ZCode launch, and the new product extends that positioning into the software development tooling market [1].


Sources

  1. The Decoder — Zhipu AI launches ZCode to challenge Claude Code and OpenAI Codex at a fraction of the cost

This article was drafted with AI from the cited sources and checked against them before publication. Spot an error? Let us know.

The open-source proxy exploits Anthropic's flat-rate image pricing to pack thousands of characters of context into a single low-token PNG.

A new open-source tool called pxpipe converts large text inputs for Claude Code into compact PNG images, cutting session token costs by 59 to 70 percent on average, according to its developer [1].

The savings matter because many Claude Code sessions burn through tokens on static, repetitive content — system prompts, tool documentation, and older chat history — that gets re-sent with every request [1]. pxpipe targets exactly that overhead.

How the pricing gap works

The technique exploits a structural difference in how Anthropic charges for text versus images [1]. Text is billed at roughly one token per character, while images are billed at a fixed token count based on pixel dimensions, regardless of how much information they contain [1]. That means dense content rendered as an image yields about 3.1 characters per image token — far better than the one-to-one text rate [1].

In practice, pxpipe runs as a local proxy that intercepts outgoing Claude Code requests [1]. It renders the bulky, static portions of a request as a single densely packed PNG, while recent messages and model outputs continue to travel as plain text [1]. Developer Steven Chong demonstrated the effect with a Fable 5 session: costs dropped from $42.21 to $6.06 [1]. A concrete example from the repository shows roughly 48,000 characters of system prompt and tool documentation — about 25,000 tokens as text — compressed into a single image costing around 2,700 tokens [1].

Tradeoffs and model compatibility

The approach is not lossless [1]. Exact strings such as hashes can come back garbled when the model reads them from a rendered image, and processing is slower because the model must run each PNG through a vision encoder rather than reading text directly [1].

Model support varies [1]. By default, pxpipe works with Claude Fable 5 and GPT 5.6, both of which Chong says handle image context reliably [1]. Opus 4.7 and 4.8 misread about 7 percent of rendered images, and GPT 5.5 also performs worse with image context; both are disabled by default and require manual opt-in [1].

Feeding text to AI models as images is not a new concept — Deepseek built an optical character recognition (OCR) system that processes text documents as images and, according to its technical paper, compresses them by up to a factor of ten while retaining 97 percent of the information [1]. What pxpipe adds is a practical, drop-in proxy layer for everyday Claude Code use.

Chong noted that if the technique gains traction, AI companies could respond by raising image processing prices [1].


Sources

  1. The Decoder — Open-source tool pxpipe hides text in PNGs to cut Claude Code and Fable 5 token costs up to 70%

This article was drafted with AI from the cited sources and checked against them before publication. Spot an error? Let us know.

Thariq Shihipar argues that with Fable 5, output quality is now bottlenecked by the user's ability to surface their own blind spots — not by the model.

An Anthropic developer has published a detailed prompting guide for Claude’s latest model, Fable 5, arguing that the biggest barrier to good output is no longer the model itself but the user’s unexamined assumptions [1].

Thariq Shihipar, writing about his own agentic coding workflow, says Fable 5 is “the first model where output quality is limited by the user’s ability to clarify their ‘unknowns'” [1]. For developers building with AI agents, that reframes the core problem: less about coaxing the model and more about interrogating your own thinking before you write a single line of a prompt [1].

A Four-Category Framework for What You Don’t Know

Shihipar organizes the problem using four categories [1]. “Known Knowns” are facts already in the prompt. “Known Unknowns” are gaps you’re aware of. “Unknown Knowns” are things so obvious you’d never write them down. The most dangerous category, he argues, is “Unknown Unknowns” — things you haven’t considered at all [1].

Getting the prompt specificity wrong in either direction causes failures, according to Shihipar [1]. Too much detail and Fable 5 rigidly follows instructions even when a course correction would be better; too little and the model falls back on generic industry defaults that may not fit the task [1]. “When you don’t account for your unknowns you fail both ways,” he writes [1].

Pre-Implementation Techniques

Shihipar describes several techniques for the phase before any code is written [1]. The first is a “blindspot pass,” in which the developer explicitly asks Claude to identify unknown unknowns — particularly useful when working in an unfamiliar part of a codebase [1]. An example prompt he offers: “I’m working on adding a new auth provider but I know nothing about the auth modules in this codebase. Can you do a blindspot pass to help me figure out my relevant unknown unknowns and help me prompt you better.” [1]

For domains heavy in “Unknown Knowns” — he cites visual design as an example — Shihipar recommends generating several radically different directions as HTML artifacts to react to, rather than jumping straight to implementation [1]. He also describes “structured interviews,” where Claude asks the developer question by question about ambiguities, prioritizing questions whose answers would most affect the architecture [1].

Source code, even in a different programming language, is his preferred reference material [1]. He notes that Claude Design reads a website’s underlying code rather than a screenshot [1]. Before work begins, he has Claude produce an implementation plan focused on the parts most likely to change — data models, type interfaces, and user-facing elements — leaving mechanical refactoring for last [1].

During and After Implementation

Unknowns don’t stop surfacing once coding starts, Shihipar warns [1]. He asks Claude Code to maintain a temporary “implementation-notes.md” file tracking decisions as they’re made, so the team can learn from each attempt [1]. When unexpected edge cases appear, his rule is for Claude to pick the conservative option, log the deviation, and keep moving [1].

After implementation, he uses two additional techniques [1]. “Pitches and explainers” are summary documents that bundle the prototype, specs, and implementation notes for stakeholders [1]. “Quizzes” have Claude generate an HTML report on the changes made, followed by a quiz — Shihipar says he does not merge code until he passes the quiz without errors [1].

He illustrates the full workflow with the launch video for Fable, which he edited entirely using Claude Code despite having no prior video editing experience [1]. When the result looked flat in color grading, he did not simply ask Claude to generate variations; he first had Claude teach him what good color grading looks like, in order to surface his own unknowns before evaluating options [1].

“Every explainer, brainstorm, interview, prototype, and reference is a cheap way to find out what you didn’t know before it gets expensive to fix,” Shihipar writes [1]. He has also published a visual version of the framework on a dedicated website [1].


Sources

  1. The Decoder — Anthropic developer shares prompting tips for Fable 5 that focus on finding your own blind spots first

This article was drafted with AI from the cited sources and checked against them before publication. Spot an error? Let us know.

As AI coding tools multiply individual engineer output, companies like Anthropic are hiring more product managers—not fewer—to keep up with accelerated software delivery.

Anthropic recently directed its growth team to hire more product managers in response to a structural shift caused by its own AI coding tool, Claude Code, which has effectively tripled the output of its engineering organization, according to industry coverage cited by VentureBeat. [1]

The dynamic illustrates a broader pattern: as AI-assisted development accelerates how fast code gets written, the constraint in software delivery has moved from implementation to product decision-making—determining what to build in the first place. [1]

The traditional ratio of product managers to engineers, historically around 1:8, now functions closer to 1:20 at companies that have deployed agentic workflows in production, because each engineer ships more per day. [1]

LinkedIn has responded to a similar pressure by replacing its associate product manager track with a “Product Builder” program that trains generalists across product, design, and engineering. [1]

VentureBeat contributor Ishan Gupta, a software engineer at Amazon, traces the shift through five phases of AI tooling adoption. [1] The first phase, which he calls the Stack Overflow era, ran from roughly 2014 to late 2022; new monthly questions on Stack Overflow have since dropped approximately 77% from their November 2022 levels, coinciding with the launch of ChatGPT. [1]

Subsequent phases saw AI models move from browser tabs into integrated development environments (IDEs) such as Cursor and Claude Code, which gave models access to full code repositories and reduced reliance on senior-engineer escalation. [1]

In the spec-driven phase spanning 2025 to 2026, larger context windows enabled work that previously required tickets, design documents, and multi-week sprints to be compressed significantly. [1] An Amazon Web Services (AWS) engineering team reportedly completed an 18-month rearchitecture project—originally scoped for 30 engineers—with 6 people in 76 days. [1] Separately, Amazon’s Kiro IDE team reportedly compressed feature builds from two weeks to two days using spec-driven workflows. [1]

In April 2026, Anthropic shipped Claude Code Routines, described as scheduled, persistent agents that can run on a cadence, on a webhook, or overnight. [1] The release effectively added an orchestration dimension to the engineering role, with engineers spinning up automated agent workflows and reviewing resulting pull requests in the morning. [1]

A third-party tool called OpenClaw, which extended similar capabilities from the open-source side, was briefly suspended by Anthropic in April before partial reinstatement. [1]

The 2025 Stack Overflow developer survey found 84% of developers using AI tools, while 46% said they do not trust the output—up sharply from 31% the year before. [1] That gap between high adoption and low trust is where code review skills have become most critical, according to the analysis. [1]

The piece argues that foundational engineering knowledge—operating systems, networking, concurrency, query optimization—has become a higher-leverage skill rather than a lower-priority one, because the blast radius of undetected errors in AI-generated code is larger at scale. [1] “The agent that wrote 70% of the code in a modern repo cannot reliably tell anyone where its assumptions about thread safety, memory ownership, or transaction isolation diverged from the runtime,” the article states. [1]

For engineers, the practical implication outlined is that product skills—talking to customers, generating validated ideas, and working backward from user outcomes—are no longer optional specializations but are increasingly required to fill the gap left by a PM-to-engineer ratio that has not kept pace with AI-driven output. [1]


Sources

  1. VentureBeat — Claude Code turned every engineer into three. Now companies need more product thinkers

This article was drafted with AI from the cited sources and checked against them before publication. Spot an error? Let us know.

Claude Code's new Artifacts feature lets teams turn coding sessions into shareable, auto-updating interactive web pages.

Anthropic has added Artifacts support to Claude Code, its command-line and desktop coding tool, enabling teams to convert coding sessions into live, shareable web pages — a capability previously available only in the Claude chat interface. [1]

The pages are generated from the full session context, including code, connected tools, and chat history. [1] When underlying content changes, the page updates automatically at the same URL and maintains a version history. [1]

Developers can trigger the feature by asking Claude Code for an artifact or for anything visual during a session; the tool then produces a shareable link accessible in a browser or the desktop application. [1] Anthropic lists example use cases including pull-request (PR) walkthroughs, incident timelines, license audits, and architecture overviews. [1]

On the access-control side, artifacts are private by default and visible only to authenticated members of an organization. [1] Administrators can manage access through role assignments and retention policies. [1]

The feature is currently in beta and is available to Claude Team and Enterprise customers. [1] It works with both the Claude Code command-line interface (CLI) and the desktop app. [1]


Sources

  1. The Decoder — Anthropic brings Artifacts to Claude Code, letting teams share live pages from coding sessions

This article was drafted with AI from the cited sources and checked against them before publication. Spot an error? Let us know.