Agentic software modernization with program analysis, dependency intelligence, migration planning, context-grounded code conversion, syntax-aware validation, bounded repair, and release gates.

LegacyLens analyzes a legacy repository, builds a structured representation of its codebase, creates a migration plan, performs context-grounded code transformation, validates generated source, and evaluates the resulting target project with deterministic engineering checks.

The conversion path is deliberately hybrid: deterministic repository analysis and validation remain outside the model, while the LLM is used for semantic translation and bounded repair where explicit transformation rules are insufficient.

LegacyLens is intentionally presented as a developer migration workbench, not a generic dashboard:

• Migration-oriented workspace and live workflow stages
• Before/after change exploration
• Release-readiness presentation
• Ask-the-codebase workspace
• Persistent migration/task context
• Hosted-safe Streamlit sidebar collapse/reopen behavior
• System-adaptive light/dark theme with live switching
• Failure states expose a downloadable failure report instead of leaving the user at a dead end

The task lifecycle is explicitly bound to its backend task ID so workflow progress events can update the task being polled by the UI.

Migration request
      ↓
Task created + bound
      ↓
Scanner / planning / conversion progress
      ↓
Task status endpoint
      ↓
Live migration workspace

The migration platform treats conversion as a verification loop rather than a one-shot generation step:

• LLM inference is routed through the shared Portfolio LLM Gateway using a request-scoped JWT.
• Deterministic repository analysis and validation remain outside the model.
• Generated source is sanitized and syntax-validated before it is accepted as migrated code.
• Syntax failures can trigger a bounded, targeted repair pass instead of a full regeneration.
• Failed conversion steps are propagated as migration failures rather than being counted as successful conversions.
• Post-migration quality gates determine release readiness.
• Blocked migrations are surfaced with explicit failure/review states rather than being presented as successful releases.

Legacy Repository
       ↓
Program Analysis
  AST / CTags / dependencies / stack detection
       ↓
Knowledge Base
  analysis + dependency context
       ↓
Migration Planning
  target architecture + symbol/file mapping
       ↓
Agentic Conversion
  deterministic rules + context-grounded LLM translation
       ↓
Syntax Validation
  AST / tree-sitter checks + bounded repair
       ↓
Post-Migration QA
  structural + executable validation
       ↓
Release Gate
  ready / blocked / review
       ↓
Migration artifact or failure report

• Multi-language source scanning
• AST / tree-sitter analysis
• Universal CTags symbol extraction
• File and symbol dependency analysis
• Technology/framework detection
• Knowledge-base retrieval over repository analysis and dependency context
• Planning before conversion
• Agentic code transformation with deterministic conversion guidance
• Syntax-aware generated-code validation and bounded repair
• Deterministic post-migration QA
• Release gates and repair loops
• Evidence-backed migration reports
• Ask-the-codebase retrieval over migration artifacts

LegacyLens uses the portfolio's shared LLM Gateway as its inference boundary.

Portfolio
      ↓
Create request-scoped BYOK session
      ↓
Short-lived JWT
      ↓
LegacyLens
  X-LLM-Gateway-Token
      ↓
Portfolio LLM Gateway
      ↓
Selected provider / model

LegacyLens does not use direct provider API keys for inference in the hosted portfolio path. The gateway handles provider/model selection and credential custody.

cp .env.example .env
docker compose up --build

python -m pytest -q
python portfolio_quality/quality_gate.py .

LegacyLens is a portfolio and engineering demonstration, not a turnkey enterprise migration service. Semantic equivalence remains difficult to prove automatically, so the system exposes structural signals, executable checks, release gates, and review-oriented evidence rather than pretending an LLM confidence score proves correctness.

Developer modernization workbench — program analysis, migration planning, context-grounded conversion, syntax-aware validation, bounded repair, executable checks, and release-aware migration workflows.