Code Raccoon

Quick Verdict

Code Raccoon, or 代码小浣熊, is the coding assistant in SenseTime’s Raccoon product family. Its official Chinese product page remained reachable on July 21, 2026, so it should not be described as discontinued. It works as an assistant inside VS Code and JetBrains-family IDEs rather than as a replacement editor, helping with completion, generation, explanation, correction, refactoring, and test ideas. Its clearest advantage is a Chinese-first workflow with direct mainland-China availability.

The public information is less complete than the documentation offered by larger global coding platforms. SenseTime’s current page does not expose a dependable public release number, detailed changelog, full language-by-language compatibility matrix, enterprise deployment matrix, or durable pricing table. Historical materials mention support for more than 30 languages, but buyers should verify the currently installed extension rather than treating that old figure as a permanent guarantee. Generated code and generated tests remain suggestions. Passing test output does not prove that requirements, permissions, security boundaries, or production behavior are correct, and a human must approve every merge.

Best For

• Chinese-speaking developers who want prompts, onboarding, and support material designed for the mainland market.
• VS Code or JetBrains users who need inline completion, code explanation, routine refactoring, and a first draft of tests without changing editors.
• Students and individual developers who want to evaluate a domestically operated assistant on ordinary, non-sensitive projects.
• Teams willing to begin with a least-privilege pilot and independently verify every diff, command, and test result.
• It is not the strongest fit for organizations that require an openly licensed client and model, fully offline operation, a published retention schedule, or a documented on-premises edition before evaluation. Those requirements need written confirmation from SenseTime.

Key Features

• IDE completion and generation: Uses the current file, comments, selected code, and instructions to suggest code without requiring constant browser copying.
• Code explanation and Q&A: Explains functions, errors, and common framework concepts in Chinese, which can make unfamiliar code easier to approach.
• Correction and refactoring: Suggests fixes or structural improvements for selected code. Repository-wide design changes still need an engineer to inspect the full diff.
• Test assistance: Can propose unit tests and edge cases. The assertions must be reviewed because an AI can reproduce the same mistaken assumption found in the implementation.
• IDE and language coverage: The product has historically supported VS Code, IntelliJ IDEA and other JetBrains environments, with mainstream languages such as Python, Java, JavaScript, C/C++, and Go. Confirm exact versions and less common languages in the live marketplace listing.
• Ownership and license: SenseTime operates the proprietary hosted product. The public product page does not publish an open-source license for its service, model, or complete client.

Use Cases

Use Code Raccoon to complete repetitive application code, explain a legacy function, sketch a conversion between languages, or enumerate tests for an existing implementation. It can also help a learner understand an error before reading the relevant language or framework documentation. These are reviewable, bounded tasks; they are safer than asking an assistant to redesign an entire repository in one pass.

For a company pilot, start with a small repository that contains no production credentials or regulated data. Grant access only to the files needed for the evaluation. Never paste .env files, access tokens, private keys, customer records, or unredacted production logs into chat. Record what the extension can read, require changes on a branch, run the project’s existing test suite and static checks, and have another person review the result. Any agent-like action that writes files, executes a command, opens a pull request, or changes configuration should require explicit human approval.

Pricing

The directory records Code Raccoon as freemium because an individual entry point and IDE product are available, but the official public page did not present a stable, citable pricing and quota table during this review. Old articles may quote free quotas or membership offers that no longer match the account experience. Individual users should check the limits shown after signing in and should avoid selecting the tool solely on an old price screenshot.

Enterprise buyers need more than a price. Ask SenseTime to document administrator controls, identity integration, repository boundaries, model providers, whether submitted code is used for training, data location, retention and deletion periods, support commitments, and any private deployment option. A sales statement is not a substitute for a DPA, security review, or enforceable contract.

Pros

• The official Chinese product surface remains online and is intended for direct mainland access.
• Integrates with familiar VS Code and JetBrains workflows instead of forcing an editor migration.
• Chinese code explanation and prompt handling can be more approachable for local users.
• Combines completion, Q&A, refactoring, and test suggestions in one development context.
• Suitable for a low-cost, limited pilot before an organization commits to a wider rollout.

Cons

• No clear public current release number or detailed changelog was available during review.
• Current IDE, language, framework, and quota coverage must be verified extension by extension.
• Public enterprise security, deployment, and retention documentation is limited.
• The product is proprietary, so it does not offer the auditability of an open-source, self-hosted assistant.
• Complex multi-file work, security repairs, and architecture decisions still depend heavily on experienced human review.

Alternatives

FAQ

Is Code Raccoon still available in 2026?

Yes. SenseTime’s official Raccoon site and coding product page were reachable on July 21, 2026. Verify the latest extension, sign-in flow, and quota before relying on it for a project because the site does not publish a detailed release timeline.

Which IDEs and languages does it support?

It has been offered for VS Code and JetBrains-family IDEs and has historically advertised more than 30 languages, including common choices such as Python, Java, JavaScript, C/C++, and Go. The current extension listing is the authoritative source for exact IDE versions and niche languages.

Is Code Raccoon open source?

No open-source license is published for the complete product, hosted service, or model. Treat it as a proprietary SenseTime service, not software that can automatically be audited, modified, and redistributed.

Can I connect a private repository?

Only after a security and policy review. Limit repository permissions, exclude secret files, use a non-production pilot, and confirm retention, deletion, and training terms. Domestic ownership does not remove those obligations.

Does passing an AI-generated test prove the code works?

No. A generated test may assert the wrong behavior, mirror a bug, or omit authorization, concurrency, failure, and integration paths. Review assertions, add independent cases, and run the existing CI and security checks.

What should an enterprise verify before purchase?

Verify SSO, administrator policy, audit logs, support and SLA, deployment options, model routing, code use, data residency, retention, deletion, and incident response in writing. Keep human approval for source changes and repository actions.

Bottom Line

Code Raccoon remains a real, reachable Chinese AI coding assistant, not an abandoned listing. It is most credible as an IDE companion for Chinese-speaking VS Code and JetBrains users who want completion, explanation, routine refactoring, and test ideas. Its weakness is information transparency: release history, exact compatibility, durable pricing, enterprise controls, and licensing are not documented publicly at the level needed for a high-trust deployment. Evaluate it with a small non-sensitive repository, measure suggestion quality and error rate, and retain normal source review, independent testing, secret isolation, least-privilege access, and human merge approval. If on-premises deployment or auditable licensing is mandatory, obtain a written vendor answer before adoption.