Mitos
An AI developer tool that traces how behavior moves through GitHub repositories and turns it into interactive diagrams, synchronized source code, and step-by-step explanations.
Overview
AI Systems / Code intelligence
Mitos is a full-stack code intelligence platform that turns GitHub repositories into shareable execution knowledge artifacts. A user describes the flow they want to understand, then Mitos authenticates with GitHub, pins the repository to an immutable commit, indexes source symbols and imports, runs a staged AI tracing agent, verifies the generated coordinates, and renders the result as a synchronized FlowBoard, Monaco code pane, timeline, and narrative explanation.
Technical stack
6 stack groups
Core Framework
4 instruments in this layerKey features
7 capability groups
GitHub Repository Trace Setup
5 implementation notes- 01
GitHub OAuth sign-in keeps identity and control-plane tokens encrypted at rest with AES-256-GCM; repository contents use short-lived, repository-restricted GitHub App installation tokens.
- 02
Repository and branch selection use public access or authenticated GitHub App installation access, with exact commit SHA pinning for every trace.
- 03
Custom trace parameters let users model runtime inputs such as roles, methods, payload shapes, or feature flags.
- 04
Trace and run history persists generated traces, previous parameter sets, bookmarks, reruns, and deletion workflows.
- 05
Public share links expose completed traces while keeping file and symbol access limited to the visible trace path.
Architecture notes
9 recorded decisions
- 01
Mounted a modular ElysiaJS API inside Next.js App Router and shared route types with the browser through Eden type-only imports, avoiding backend code in the client bundle.
- 02
Designed a durable Inngest trace pipeline with per-trace concurrency, cancellation, optional Redis worker locks, attempt-scoped CAS idempotency, and artifact offloading.
- 03
Built a two-phase AI tracing system that separates bounded repository exploration, behavioral evidence extraction, canonical trace planning, and structured output drafting.
- 04
Implemented universal directory relevance scoring so large repositories surface READMEs, manifests, docs, source files, entry points, MVC/API paths, and prompt-matching files before noise.
- 05
Engineered a five-layer verifier that repairs coordinates from real source text, merges overlapping steps, audits import graph connections, scores quality, and locks verified coordinates after semantic review.
- 06
Created repository intelligence infrastructure for tree syncing, import maps, path alias resolution, symbol parsing, proactive/lazy indexing, and relational BFS fallback traversal.
- 07
Synchronized FlowBoard, Monaco, timeline, playback, keyboard shortcuts, hover states, and narrative content through a trace workspace optimized for desktop and mobile workflows.
- 08
Hardened public trace sharing with trace-scoped file/symbol authorization, GitHub App installation-scoped repository reads, encrypted identity tokens, text archive fallbacks, binary rejection, and Mermaid/SVG sanitization.
- 09
Established a benchmark harness and diagnostics layer that tracks golden fixtures, anchor coverage, evidence grounding, tool behavior, stage timings, and generalization risk.
Project status
3 status records
Core Product Implemented
5 recorded items- 01
GitHub OAuth, repository/branch picker, custom trace form, model configuration checks, and trace creation flow.
- 02
Durable trace generation pipeline with repository resolution, symbol indexing, AI exploration, evidence packing, planning, drafting, verification, self-correction, and persistence.
- 03
Interactive trace workspace combining Mermaid FlowBoard, Monaco code pane, timeline, playback controls, narrative drawer, validation drawer, command palette, and keyboard shortcuts.
- 04
Trace and run history with bookmarks, parameter reuse, reruns, deletion, public share toggles, and copy-link actions.
- 05
Responsive layout with mobile Diagram/Code/Steps switching, persisted panel visibility/order, and Motion-powered panel transitions.