Release/v1.1 dsa engine#33
Merged
Merged
Conversation
- Implement circular dependency detection on adjacency list - Add percentile-based file complexity heatmap to graph nodes - Introduce orphan file detection for dead code identification - Update node state interfaces and UI legends
…nents - Extract state management, LLM streaming, and graph traversal into custom hooks - Modularize UI into dedicated layout components (Visualizer, Doctor, Risk Radar) - Isolate static configurations and types into constants.ts - Strip page.tsx down to act strictly as a layout orchestrator
- Extract changed files from existing PR analysis route - Add 'pr-blast' highlight mode to React Flow ArchitectureMap - Implement automatic tab switching and state lifting for seamless UX
…ive dashboard - **Algorithms**: Shipped zero-latency, pure math static analysis features including Articulation Points (Bridge Detection), Dead Code Reachability (BFS), Dependency PageRank, and Robert C. Martin's Instability Matrix (Ca/Ce coupling). - **Architecture Insights UI**: Built the new `ArchInsightsPanel` and `InfoTooltip` to visualize critical single-points-of-failure, unreachable code chunks, and codebase stability metrics directly over the React Flow canvas. - **Pipeline Integration**: Upgraded `ast-pipeline.ts` and `analyze.ts` types to support the new graph theory outputs without triggering external API calls. - **Command Center Dashboard**: Extracted flat dashboard list into a client-side `DashboardGrid` component. Added interactive grouping by repository, analysis count badges, and chronological sorting with formatted timestamps.
…hangedFiles comment
- Add checkout API route (/api/checkout) - Add webhook handler (/api/webhooks/lemonsqueezy) - Update usage.ts to support pro tier (100/day limit) - Update analyze route to use ratelimitPro for pro users - Update chat route to bypass limit for pro users - Update analyze-pr route to bypass limit for pro users - Add ratelimitPro to ratelimit.ts - Supabase profiles table created with plan_tier column
- Add handleUpgrade function calling /api/checkout - Update tier names: Student→Intern, Architect→Specialist, Surgical→Chief Surgeon - Specialist tier now live with ₹99/mo and amber styling - Add loading state and login redirect for unauthenticated users - Remove Coming Soon badge from Specialist tier
- Free tier: context 20k -> 6k chars, max_tokens 1024 -> 512 - Pro tier: context 14k chars, max_tokens 1024, deepseek-r1 model - Reduces token usage by ~70% for free tier chat
Amber color tokens felt out of place against the dark technical aesthetic. Replaced all amber instances in RiskDashboard with the existing slate/white/transparent vocabulary used throughout the component. - Pro banner: amber border/bg → white/5 glass surface - Upgrade button: amber fills → slate-300/white hover - Locked Auto-Patch button: amber tint → desaturated slate-500 - FileCode icon: amber-400 → slate-300 with white hover - Traffic light middle dot: amber-500 → slate-500
Amber color tokens felt out of place against the dark technical aesthetic. Replaced all amber instances in RiskDashboard with the existing slate/white/transparent vocabulary used throughout the component. - Pro banner: amber border/bg → white/5 glass surface - Upgrade button: amber fills → slate-300/white hover - Locked Auto-Patch button: amber tint → desaturated slate-500 - FileCode icon: amber-400 → slate-300 with white hover - Traffic light middle dot: amber-500 → slate-500
- fix: dynamic analysis_version query instead of hardcoded v3 - fix: separate visited sets for upstream/downstream BFS traversal - fix: node ID collision in mermaid sanitizer (path-aware) - fix: opacity replaced with stroke-width in mermaid styles - fix: JSON.parse wrapped in try/catch in cache and groq-debug - fix: requiresRuntimeCheck narrowed to specific runtime patterns - fix: at async removed from framework noise filter in stack-parser - fix: findMatchingFile disambiguates same-name files by path score - perf: parallel file fetching via Promise.allSettled - perf: reverse graph memoized via WeakMap - perf: BFS queue uses head pointer instead of array.shift() - refactor: rename gemini-debug to groq-debug with backward alias - refactor: confidence removed from AI prompt, heuristics only - remove: dead code generateErrorGraph and extractAllCrashNodes - add: cache invalidation via invalidateRepoCache() - add: server-side rate limiting with 429 response
The React Flow node/edge renderer — wherever node styles and badges are applied The right analysis panel — wherever CIRCULAR DEPS / ORPHANS / NERVE CENTER are rendered The Risk Radar tab — its current state
…d as prop - Add computePageRank() helper using damped iteration (20 passes, 0.85 factor) - Normalize scores to 0-100 range for consistent tier thresholds - Add computedPageRankScores useMemo that falls back to internal computation - Replace all pageRankScores references with computedPageRankScores - Nerve Center panel now works out of the box without parent wiring
- Logged-in users now get a fresh DB row on every analysis (cached or not) so the dashboard always reflects the latest activity with correct timestamp - RAG storage still skipped for cached results to avoid redundant processing - Anon users retain original behavior (insert only on fresh analyses) Fixes: dashboard showing stale May 29 dates after re-analyzing existing repos
- Switch all routes from llama-3.3-70b-versatile to llama-3.1-8b-instant (5x higher token limit: 100K -> 500K TPD, same free tier) - Cache AI response in analyses.ai_response column after first run so repeat analysis of same repo costs 0 tokens - Tee live stream to store response without breaking streaming UX - Cap fileContents to top 20 files in gemini.ts to prevent token blowout on large repos (was sending all files unbounded)
…ty, and cfg builder
Contributor
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
🧠 Architecture Analyzed by CodeAutopsyLatest commit scanned: 
|
There was a problem hiding this comment.
Pull request overview
Release v1.1 of the DSA engine: adds new graph algorithms (betweenness centrality), a control-flow-graph static analyzer, repo-type detection and graph-quality assessment, a depth-based file selector, a semantic analysis cache, several performance/perf-correctness fixes in ArchitectureMap, and switches the streaming LLM from Groq Llama-3.3-70B to Cerebras gpt-oss-120b (with smaller fallback models for non-streaming endpoints).
Changes:
- New analysis modules:
lib/algorithms/betweenness.ts,lib/cfg-builder.ts,lib/file-filter.ts,lib/analysis-cache.ts, plusgetFilesByDepthinlib/dependency-graph.ts, all wired intorunAstPipelineand surfaced throughPipelineResult/RepoData. - Pipeline hardening: PageRank uses
reduceinstead of spread to avoid call-stack blowups, source nodes are preserved when all targets are config files, local uploads now run language detection, andMAX_FILES_TO_FETCHis raised to 60. - UI:
ArchitectureMapmemoizes nodes/sidebar, splits layout vs. score memos, renders betweenness severity rings/badges, andVisualizerPanel/app/analyze/page.tsxforward the newbetweennessScoresprop; analysis-version bumped to 14 and analyze caching gains auser_idfilter and per-user insert path.
Reviewed changes
Copilot reviewed 19 out of 20 changed files in this pull request and generated 6 comments.
Show a summary per file
| File | Description |
|---|---|
| package.json / package-lock.json | Bump openai to ^6.40.0; add @babel/parser & @babel/types as devDeps for the new CFG builder. |
| lib/types/analyze.ts | Add betweennessScores, cfgFindings, cfgSummary to RepoData. |
| lib/pipeline/ast-pipeline.ts | Integrate file-filter, betweenness, CFG batch, depth-based selection, graph-quality metadata; misc cleanups and commented bug fixes. |
| lib/dependency-graph.ts | Add getFilesByDepth BFS selector and tidy whitespace. |
| lib/file-filter.ts (new) | Multi-layer skip / graph-node / repo-type / graph-quality filters. |
| lib/cfg-builder.ts (new) | Babel-based detector for unreachable code, missing error handling, infinite loops. |
| lib/algorithms/betweenness.ts (new) | Brandes' betweenness centrality with severity mapping & colors. |
| lib/analysis-cache.ts (new) | Helpers for HEAD-SHA fetch, staleness check, and cache invalidation. |
| lib/gemini.ts | Switch streaming model/provider to Cerebras gpt-oss-120b; cap file context to 20. |
| lib/debug/groq-debug.ts | Downgrade debug model to llama-3.1-8b-instant. |
| components/ArchitectureMap.tsx | Memoize node/sidebar, split layout/score memos, render betweenness ring/badge. |
| components/DirectoryTreeVisualizer.tsx | Move nodeTypes to module scope; minor cleanups. |
| components/analyze/VisualizerPanel.tsx | Pass betweennessScores through; show legend chip. |
| app/analyze/page.tsx | Forward betweennessScores to the visualizer. |
| app/api/analyze/route.ts | Bump ANALYSIS_VERSION, scope cache by user_id, split logged-in vs. anonymous insert/RAG paths. |
| app/api/v1/analyze/route.ts | Bump ANALYSIS_VERSION to 14. |
| app/api/ai/route.ts | Add per-user ai_response cache with tee'd stream write-back. |
| app/api/analyze-pr/route.ts, app/api/generate-test/route.ts | Switch to llama-3.1-8b-instant. |
| lib/tests/*.test.ts, lib/algorithms/tests/betweenness.test.ts | Tests for file-filter, depth selector, CFG builder, and betweenness. |
Comments suppressed due to low confidence (1)
app/api/analyze/route.ts:223
- For anonymous requests
userIdisundefined, so this becomes.eq("user_id", ""). If theuser_idcolumn isuuid, Postgres will reject the empty string and the cache query will return an error (logged as a warning and treated as a miss), so anonymous users will never get a cache hit. Additionally, even if the column accepts empty strings, anonymous rows are inserted withuser_id: null(line 281), which.eq("user_id", "")will never match. Use.is("user_id", null)whenuserIdis absent.
.eq("user_id", userId ?? "")
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
| .select("commit_sha, created_at") | ||
| .eq("repo_url", repoUrl) | ||
| .eq("analysis_version", ANALYSIS_VERSION) | ||
| .eq("user_id", userId ?? "") |
Comment on lines
+104
to
+111
| for (const stmt of body.body) { | ||
| if (stmt.type === "TryStatement") { | ||
| hasTryCatch = true; | ||
| } | ||
| if (JSON.stringify(stmt).includes('"type":"AwaitExpression"')) { | ||
| hasAwait = true; | ||
| } | ||
| } |
Comment on lines
+294
to
+297
| const eligible = fileContents.filter((f) => { | ||
| const ext = "." + f.path.split(".").pop(); | ||
| return analysableExtensions.has(ext); | ||
| }); |
Comment on lines
7
to
10
| const baseGroq = new OpenAI({ | ||
| apiKey: process.env.GROQ_API_KEY!, | ||
| baseURL: "https://api.groq.com/openai/v1", | ||
| apiKey: process.env.CEREBRAS_API_KEY!, | ||
| baseURL: "https://api.cerebras.ai/v1", | ||
| }); |
Comment on lines
+72
to
+73
| // NOTE: NON_JS_REPO removed — all languages are now supported. | ||
| // Sparse graphs show folder structure + metrics instead of failing. |
Comment on lines
+653
to
+656
| const betweennessResult = computeBetweenness(dependencyGraph, true); | ||
| const betweennessScores: Record<string, number> = Object.fromEntries( | ||
| betweennessResult.scores, | ||
| ); |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.