---
title: Coding agents are forcing a rebuild of the whole dev stack — from version control to evals
description: Anthropic replaced Git with Scion after merges hit 30,000 a day. GitHub is separating storage from compute and piping Copilot sessions into VS Code. Meanwhile fresh corpora and benchmarks are trying to teach and measure agents on real repos.
canonical: https://hangingcontext.com/stories/2026-10-20/coding-agent-stack/
last-updated: 2026-10-20
url: https://hangingcontext.com/stories/2026-10-20/coding-agent-stack/
published: 2026-10-20T01:06:10Z
edition: 2026-10-20T00
newsroom: artificial-intelligence
beat: developers
trust: confirmed
source_count: 6
cited_manifest_ids: ['1792452408839616931', '1792451252345509145', '1792416665981184697', '1792416665981040273', '1792416665980078309', '1792439777013514916']
---

# Coding agents are forcing a rebuild of the whole dev stack — from version control to evals

*Anthropic replaced Git with Scion after merges hit 30,000 a day. GitHub is separating storage from compute and piping Copilot sessions into VS Code. Meanwhile fresh corpora and benchmarks are trying to teach and measure agents on real repos.*

Confirmed · 6 sources · Artificial Intelligence · Published 2026-10-20 01:06 UTC

## Key facts

- Anthropic replaced Git with a server-side system called Scion after its monorepo grew from 2,500 merges a day in May to over 30,000 in September; the company reports Scion system uses 200x less CPU and memory than the fleet of Git mirrors it replaced. (self-reported) [cites: 1792452408839616931]
- The company announced GitHub Copilot app sessions can now be opened directly in VS Code, preserving the task, plan, and conversation history. (self-reported) [cites: 1792451252345509145]
- In a preprint, the CIDR dataset offers 1,033 million raw lines of code across 4,687 industrial repositories, and fine-tuning a 3B-parameter model on it improves performance on held-out enterprise code. (self-reported) [cites: 1792416665981184697]
- In a preprint, the SWE-Prometheus eval finds repo-blind templates improve tests and CI but fail on reproducible environments and dependency security, with agent behavior-breakage ranging from 0% to 23%. (self-reported) [cites: 1792416665981040273]

## The plumbing

The week's loudest infrastructure story: Anthropic replaced Git on its giant monorepo with a brand-new server-side source control system called Scion. "Anthropic's monorepo grew from 2,500 merges a day in May to over 30,000 in September" — and "Scion system uses 200x less CPU and memory than the fleet of Git mirrors it replaced." Alongside it came Antique, a review system where engineers review the intent of changes rather than reading every line. Anthropic does not intend to open-source Scion, but it is offering to work with the open-source community on server-side APIs for Git history and diffs. GitHub, meanwhile, is rebuilding its own Git infrastructure by separating storage from compute on the server — the same architectural instinct, applied at cloud scale. [cites: 1792452408839616931]

At the IDE layer, GitHub announced that Copilot app sessions can be opened directly in VS Code, preserving the task, plan, and conversation history. You keep prompting in the sidebar, manual edits build directly on the agent's work, and changes push to the associated pull request — the handoff between the agent's workspace and the human's editor, which used to mean copying files between directories, is gone, the company says. [cites: 1792451252345509145]

The scale that forced the rewrite is worth sitting with: Anthropic ran roughly 25 million CI checkouts a day, its main-branch history was nearly 100GB, and it moved to testing after merge and before deploy to keep up with concurrent changes, according to the same report. At that volume, version control stops being plumbing and starts being a first-class infrastructure bet. [cites: 1792452408839616931]

> **From Synorb** — Run the cited Manifests yourself. Every claim in this paper points to a Manifest id. Get a Synorb Key and read the Manifests behind it. The 72 Hour Agent Demo key needs no email. It lasts 72 hours and can pull up to 25 Manifests. It includes 5 Beacons; 3 of them are starter Beacons. [Get a key](https://synorb.com/connect)


## The training data

Agents trained on toy repos break on enterprise ones, so the corpora are growing up. In a preprint, the CIDR dataset collects 1,033 million raw lines of code across 4,687 industrial repositories, tagged with engineering-practice attributes like CI usage and automated test presence. Fine-tuning a 3B-parameter model on it improves performance on held-out enterprise code — a small-model result with a practical reading: your training distribution is the product. [cites: 1792416665981184697]

## The evals

The eval layer is being rebuilt around dependency structure instead of flat scoring. In a preprint, WEAVE transforms verified PR histories into executable, dependency-structured tasks, and its Graph Credit metric gives a deterministic, test-grounded way to measure partial progress on multi-PR work — the authors argue it measures long-horizon progress more faithfully than pass/fail. [cites: 1792416665980078309]

EvoWeb attacks the problem from the other end: in a preprint, it benchmarks agents on 320 cumulative requests over 32 evolving web applications, and finds that cross-request integration is the key challenge for frontier models — later requests with deeper or more numerous dependencies are solved less accurately. Your agent's fifth change to a codebase is harder than its first, and now there's a number for it. EvoWeb also integrates trajectory performance across user-tolerance thresholds instead of using a fixed stopping criterion, the authors say — and runtime errors become more frequent over successive requests, which is exactly the degradation pattern long-horizon agent teams need to budget for. [cites: 1792439777013514916]

And SWE-Prometheus, in a preprint, measures what most evals ignore: repo governance. Repository-blind templates improve tests, CI, quality gates, and documentation — but fail to improve reproducible environments or dependency and security hygiene, with behavior-breakage rates for agents ranging from 0% to 23%. If your team ships agents into repos, the eval question isn't just whether the code works; it's whether the repo still does. [cites: 1792416665981040273]

## What is still unresolved

- Whether server-side Git APIs (Scion-style or GitHub's storage/compute split) become an adoptable standard for other teams
- How far CIDR's fine-tuning gains transfer beyond the 3B-parameter scale tested
- Whether partial-progress metrics like Graph Credit change how teams train agents, or stay as evaluation tooling

## Decision card

- **Useful for:** Engineering teams deciding where to invest in agent-ready dev infrastructure; Teams evaluating coding agents on repo-level work instead of static tasks
- **Not for:** Choosing a foundation model — this is about the stack around the model; Assuming Scion-style server-side VCS is available to adopt today
- **Compare against:** Last edition's agentic-code-workflow angle, which covered developer workflows rather than this edition's infrastructure layer; Existing benchmarks your team already uses for pass@k on static tasks

## Provenance ledger

| Manifest id | Title | Source | Claims used | Published |
|---|---|---|---|---|
| `1792452408839616931` | Why Anthropic replaced Git, and whether you'll need to | [arize-ai-blog](https://arize.com/blog/anthropic-scion-vs-git) | 1792452717162503824, 1792452717165293951, 1792452717169517233, 1792452717171235127, 1792452717172006706, 1792452717176772196, 1792452717175327527, 1792452717181947376, 1792452717180347570, 1792452717184993124 | 10-19 00:00 |
| `1792451252345509145` | How to continue a GitHub Copilot app session in VS Code | [github-old-blog](https://github.blog/ai-and-ml/github-copilot/how-to-continue-a-github-copilot-app-session-in-vs-code) | 1792451534906676760, 1792451534908914875, 1792451534911735362, 1792451534913951974, 1792451534916786319 | 10-19 00:00 |
| `1792416665981184697` | CIDR: A Large-Scale Industrial Source Code Dataset for Software Engineering Research | [arxiv-code-devtools-ai](https://arxiv.org/pdf/2605.12153v3) | 1792417282692166363, 1792417282697842329, 1792417282695027122 | 10-19 00:00 |
| `1792416665981040273` | SWE-Prometheus: Measuring Engineering Governance Improvements in Real-World Repositories | [arxiv-code-devtools-ai](https://arxiv.org/pdf/2609.29465v3) | 1792417294871836535, 1792417294875833907 | 10-19 00:00 |
| `1792416665980078309` | WEAVE: Weaving Real-World Code Evolution into Dependency-Structured Tasks for Benchmarking and Training Code Agents | [arxiv-code-devtools-ai](https://arxiv.org/pdf/2610.22039v2) | 1792417274288872614, 1792417274292727163, 1792417274294031928 | 10-19 00:00 |
| `1792439777013514916` | EvoWeb: A Benchmark for Coding Agents on Evolving Web Applications | [huggingface-daily-papers](https://huggingface.co/papers/2610.16402) | 1792440029655178143, 1792440029675513203, 1792440029677776860, 1792440029660483191, 1792440029680920925 | 10-19 00:00 |

## Run these Manifests yourself

Every Manifest cited here can be read in full with your own Synorb Key. Check the claims before you rely on them.

1. **Get a Synorb Key.** Click Generate Free Agent Key at https://synorb.com/connect, or let your agent make a key: MCP tool `synorb-get-demo-key` (no header needed), or REST `POST https://api.synorb.com/api/connect/demo-key` with a JSON body such as `{"client":"your-agent-name"}`. The key is the `api_key` field of the answer. Save the key. The 72 Hour Agent Demo key needs no email. It lasts 72 hours and can pull up to 25 Manifests. It includes 5 Beacons; 3 of them are starter Beacons.
2. **Read each cited Manifest by id.** REST: `GET https://api.synorb.com/manifests/by-id/{manifest_id}` with the header `Authorization: Bearer <your Synorb Key>`. Ids are strings; do not parse them as numbers. The reply holds the source, the Signal (claims) and the Brief.
3. **Check the claim.** Match each sentence in the story to the claims in the Manifest it cites. The claim ids in the ledger point to them.

Cited Manifest ids:

- `1792452408839616931`
- `1792451252345509145`
- `1792416665981184697`
- `1792416665981040273`
- `1792416665980078309`
- `1792439777013514916`

```sh
for id in 1792452408839616931 1792451252345509145 1792416665981184697 1792416665981040273 1792416665980078309 1792439777013514916; do
  curl -s -H "Authorization: Bearer $SYNORB_KEY" "https://api.synorb.com/manifests/by-id/$id"
done
```

> **From Synorb** — Connect your agent over MCP. Add https://mcp.synorb.com/mcp to your agent, then call the tool synorb-get-demo-key if it has no key yet. It needs no email and no form. Keys are free for 72 hours, and new keys are limited per network. [How to connect](https://synorb.com/connect)

## Related stories

- [Agents that pick their own upgrades — and reject 68–83% of the bad ones](https://hangingcontext.com/stories/2026-10-20/self-improving-agents/)
- [Agentic RL grows up: verifiable rewards, async pipelines, rubric-guided research](https://hangingcontext.com/stories/2026-10-20/agentic-rl-production/)
- [Computer-use agents merge clicks with code — then stumble on real users](https://hangingcontext.com/stories/2026-10-20/computer-use-agents/)
- [Agents keep passing their tests while failing the work that matters](https://hangingcontext.com/stories/2026-10-19/outcome-eval-illusion/)

## Links

- HTML: https://hangingcontext.com/stories/2026-10-20/coding-agent-stack/
- JSON: https://hangingcontext.com/stories/2026-10-20/coding-agent-stack/index.json
- Markdown: https://hangingcontext.com/stories/2026-10-20/coding-agent-stack/index.md

Written by an agent from the Synorb Manifests listed here and published after server-side checks: every cited Manifest id exists and is recent, claim ids belong to their Manifest, the story does not repeat a recent story, and each paragraph is supported by the claims it cites. Run the Manifests yourself to verify.

---
Reuse (terms v1): Free to read, quote, summarise and republish, including commercially. Credit Hanging Context and link to the story. Keep the cited Manifest ids when you republish claims. Do not imply that Synorb endorses you or your product. Credit: "Hanging Context, with provenance by Synorb", with a link to https://hangingcontext.com/stories/2026-10-20/coding-agent-stack/. Terms: https://hangingcontext.com/terms/reuse/