Open sourceTypeScript · ≈228k stars · MIT
opencode
The open source coding agent — terminal-first, model-agnostic.
OpenCode is an open-source coding agent with a terminal UI, a desktop app (beta) and an SDK, and works with any model provider. Version 2 adds experimental desktop plugins, and since v2.3 subagents run in the background by default; the GitHub Action was removed in v2.2.
Pros & cons
Each point cites its sources and says what kind of evidence backs it. A second reviewer tried to refute every claim before it was kept.
Pros
- User reportsWorks with any provider or model, with LSP integration and configurable subagents that can each use a different model.[1][2]
- User reportsUsers measure its prompt overhead as much leaner than Claude Code's (about 7k vs 33k tokens), and it has a plugin ecosystem.[1][2]
- User reportsA popular fallback and local-model harness when subscription limits run out; 2.0 added background tasks.[1][2]
- MetadataMIT-licensed, about 223k stars, with near-daily releases (v1.19.5 on Oct 28, 2026).[1]
Cons
- User reportsIn October 2026 the OpenCode team disclosed unauthorized access to a legacy system that may have exposed BYOK provider keys saved in hosted workspaces.[1][2]
- User reportsUsers question its security posture: no built-in sandbox, and title generation sent prompts through the Zen gateway by default.[1][2]
- User reportsUsers say 2.0 moved tool calls into a background process, which silently broke their sandbox wrappers, and that RAM use is high.[1][2]
- MetadataThousands of open issues and PRs, and some users get better results from other harnesses with the same model.[1][2]
- Best for
- Developers who want a model-agnostic terminal agent with low prompt overhead and per-subagent model choice.
- Watch out for
- Run it inside your own sandbox and set your own small model for auxiliary calls, so prompts don't leave your machine by default.
Assessed Oct 29, 2026 · adversarially reviewedSuggest a correction
People
Engineer · Anomaly
Core builder, opencode
Open harnesses
Anomaly
Co-founder & CEO
Open harnesses
Engineer
Open harnesses
Anomaly
Co-founder
Harness engineering
Activity
- Last push
- · today
- Latest release
- v2.3.2
· 3d ago - Last 90 days
- 20+ releases
- · 3d ago
Highlights GPT-6.2 Prompt Caching: GPT-6.2 models don't cache a system prompt sent in the Responses instructions field, so every step re-billed it as uncached input. OpenCode now sends it as a leading developer message on every Responses route, including OpenAI (API key and…
- · 3d ago
Highlights Timed-Out Shell Commands Keep Running: A shell command still running when its timeout expires now moves to the background instead of being killed, so long builds and test runs finish and report back. A foreground timeout: 0 now means the default 120 seconds. (#63699)…
- · 4d ago
Desktop Improvements OpenCode 1 now offers the upgrade to OpenCode 2.0. OpenCode 2 downloads in the background and installs only when you choose: Restart to upgrade, Upgrade on quit, or Later. Your sessions, projects, and settings are kept. Plugins made for OpenCode 1 need an…
process-daemon-v0.6.0process-daemon v0.6.0
· 5d agoopencode-pty 0.6.0 (protocol 9) from packages/process-daemon, published to npm as @opencode/pty@0.6.0.
- · 8d ago
Highlights Background Subagents by Default: Subagents now return immediately, continue in the background, and report back whenever they become idle. You can keep working in the parent session while multiple subagents run concurrently. (#63382) Quieter Transcripts by Default:…
Checked against GitHub on Nov 21, 2026. Release notes are shortened to plain text.
In the news
Open Source · TechCrunch AI · Oct 19, 2026
OpenCode is an open source AI coding agent that's become a popular alternative to Claude Code and Cursor. But hackers got into part of its online platform and stole some users’ API keys, and the AI tools it turned to for help kept refusing to look at the malicious code.
In research (28)
- Is This Code Mine? Student Ownership When AI Agents Write the Spec and the CodeVerdict: Incremental
Nov 16, 2026 · OpenCode was declared as an agent in the EduQuest project.
- Harness Leverage Across Model Size: A Ladder of Five Local Agent Harnesses from 0.75B to 4.7BVerdict: Useful
Nov 16, 2026 · Baseline product harness (v1.18.23); scores 0.000 at 3.7B under the deployment reading.
Nov 16, 2026 · Single agent harness used to rank the five LLMs in Section 4.
- Hidden in Plain Sight: an Empirical Study about The Instruction Stacks of Coding AgentsVerdict: Important
Oct 27, 2026 · Captured agent, the only one whose request shrank over time.
- Spine: Compiling Subtask Trees into Structured Context for Long-Horizon Coding AgentsVerdict: Important
Oct 27, 2026 · Open-source coding harness where the same context-compilation approach would apply.
Oct 27, 2026 · The harness used for GLM 5.3, DeepSeek V4 Pro, Kimi K3 and HY4 preview.
Showing the 6 newest of 28 papers.
Neighbours on the workbench
Shell · ≈329k stars · MIT
mattpocock/skills
Small, composable agent skills for real engineering work.
Maintained by Matt Pocock (AI Hero)
MDX · ≈323k stars · MIT
obra/superpowers
An agentic skills framework and software methodology that works.
Maintained by Jesse Vincent (obra)
JavaScript · ≈308k stars · MIT
affaan-m/ECC
Agent harness performance optimization system: skills, instincts, memory and security for coding agents.
Maintained by Affaan Mustafa
Connections
Relations from the Harness Hub knowledge graph, as charted in the State of Harness reports.
Maintained by · 1
- Anomaly (opencode)Startup
Similar entries · 5
- Anomaly (opencode)Startup
- CodexProject
- CrushProject
- Grok BuildProject
- PlandexProject
Sources
Verified Nov 16, 2026
