# HKUDS/DeepCode

"DeepCode: Open Agentic Coding (Agent Harness & Loop Engineering & Multi-Agent Orchestration)"

Repository: https://github.com/HKUDS/DeepCode
Canonical: https://ross.abutalabs.com/products/deepcode
Homepage: http://arxiv.org/abs/2512.07921
Language: Python
License: MIT
License Family: permissive
Topics: agentic-coding, llm-agent, harness-engineering
Last push: 2026-08-25T13:50:25+00:00

## Health v2 (maintenance only)
Score: 85/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 99, release rhythm 97, longevity 34
- inputs: {"age_days": 476, "days_push": 8, "days_rel": 21, "gap_med": 8.0, "n_releases_24m": 15}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 16434, forks 2144 (observed 2026-08-28T04:11:16.425689+00:00)

## What it is
DeepCode is an open agentic coding framework from HKUDS that orchestrates multi-agent LLM pipelines to autonomously generate production-grade codebases, notably reproducing scientific papers as working code. It uses blueprint distillation, stateful code memory, retrieval-augmented knowledge injection, and closed-loop error correction to manage LLM context limits.

## Use cases
- reproduce a research paper as runnable code automatically
- generate a codebase from a document or specification
- orchestrate multiple LLM agents for coding tasks
- benchmark autonomous coding agents against PaperBench
- build custom agent harnesses with loop engineering

## When to choose
- you need autonomous paper-to-code or document-to-codebase synthesis
- you want an open-source alternative to commercial coding agents like Cursor or Claude Code
- you are researching multi-agent orchestration and context management for code generation

## When to avoid
- you need a lightweight autocomplete-style code assistant in your editor
- your task is simple single-file code generation
- you cannot provide LLM API access or lack compute for multi-agent runs

## Facets
- artifact type: framework
- maturity: active
- function: agent-framework, llm-inference, rag, machine-learning
- domain: artificial-intelligence, large-language-models, developer-tools
- platform: python, cli, cross-platform
- tags: agentic-coding, multi-agent-orchestration, code-generation, paper-to-code, harness-engineering, autonomous-agents, ai-agents, automation

## Member repositories
- HKUDS/DeepCode (main) score 85

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:11:16.425689+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:05:06.564707+00:00, confidence not recorded.
  - readme: https://github.com/HKUDS/DeepCode (fetched 2026-08-28T04:11:16.425689+00:00, sha e155f84a1f67)
  - homepage: http://arxiv.org/abs/2512.07921 (fetched 2026-08-29T08:02:24.070114+00:00, sha 68c1f8660069)
  - site_page: https://info.arxiv.org/about/donate.html (fetched 2026-08-29T08:02:24.079808+00:00, sha cca9c3a11c56)
  - site_page: https://info.arxiv.org/about/ourmembers.html (fetched 2026-08-29T08:02:24.083487+00:00, sha 47cbc55ff1de)
  - site_page: https://info.arxiv.org/about (fetched 2026-08-29T08:02:24.085359+00:00, sha a1f16f915a9a)
  - site_page: https://info.arxiv.org/labs/index.html (fetched 2026-08-29T08:02:24.081858+00:00, sha b14a8d05a0ec)
- Data as of 2026-08-30T08:39:29.467469+00:00.