Open Source
Public code, prompts and rules
Decoder is an open study experiment: code, analysis prompts and rules live in the public repository.
What's open
Source code
The whole application, frontend and backend, under MIT.
Analysis prompts
The instructions we give to LLMs to explain, audit quality and security — editable by the community.
Rules and scoring
Finding severity, analysis kinds, aggregation criteria: all in the clear in the repository.
Translations
Three official languages today, expandable with a PR for each new language.
Design system
Semantic tokens and accessible components in src/styles.css and src/components/ui.
Optional local inference
When Ollama or LM Studio is selected, AI inference runs on your localhost and the file body is not sent to external providers. Uploaded files remain in the server's storage, accessible only to you.
Public roadmap
Next areas of work. Discussed in the open, editable by the community.
Already shipped: Knowledge Hub + Engine, folder-scoped repo chat, cross-session persistence, installable PWA, EU/India SEO clusters, 2FA password reset, cleanup endpoint hardening, 20+ static-analysis languages.
- R01GitHub OAuth login and private-repo import with user PAT (in progress).
- R02VS Code / JetBrains companion that invokes Decoder during PR review (in progress).
- R03Deeper analyses: architecture, dependencies, performance, maintainability (next).
- R04Diff-aware suggested inline comments, exportable as a PR review (next).
- R05MCP server + shared community rule library for AI-generated code smells and block-level provenance (design).
Extend the guardrail
Analysis prompts live in src/lib/analysis-prompt.ts. Add a new kind, write the instructions, open a PR: the community reviews it, we refine it together.
See the prompt fileWho builds it
Decoder is maintained by anonymous contributors. Your issue, your PR or your translation is part of the guardrail.