roadmap_version: '2.0'
github_enabled: false
github_repo: paiml/verificar
roadmap:
- id: VERIFICAR-040
github_issue: null
item_type: task
title: Add C grammar for decy transpiler
status: completed
priority: high
assigned_to: null
created: 1970-01-01T00:00:00Z
updated: 1970-01-01T00:00:00Z
spec: null
acceptance_criteria:
- CGrammar struct with tree-sitter-c integration
- CEnumerator for exhaustive generation
- Support pointer arithmetic, struct declarations, preprocessor directives
- 80+ tests covering C language features
phases: []
subtasks: []
estimated_effort: 8-12 hours
labels:
- grammar
- c
- decy
notes: Priority for CPython, Git, SQLite transpilation targets
- id: VERIFICAR-041
github_issue: null
item_type: task
title: Add TypeScript grammar for decy target
status: completed
priority: medium
assigned_to: null
created: 1970-01-01T00:00:00Z
updated: 2025-11-30T21:25:09.726162633+00:00
spec: null
acceptance_criteria:
- TypeScriptGrammar with tree-sitter-typescript
- TypeScriptEnumerator for type-safe generation
- Support interfaces, generics, async/await
- 60+ tests
phases: []
subtasks: []
estimated_effort: 6-8 hours
labels:
- grammar
- typescript
notes: null
- id: VERIFICAR-042
github_issue: null
item_type: task
title: Add Ruchy grammar for ruchy language
status: completed
priority: low
assigned_to: null
created: 1970-01-01T00:00:00Z
updated: 1970-01-01T00:00:00Z
spec: null
acceptance_criteria:
- RuchyGrammar for ruchy language syntax
- Support fn/fun, let/var, struct, enum, impl, trait
- Actor model (actor, spawn, send, receive, ask)
- Effect system, pipeline operators, optional chaining
- 81 tests covering Ruchy language features
phases: []
subtasks: []
estimated_effort: 4-6 hours
labels:
- grammar
- ruchy
notes: Lower priority - ruchy is self-hosting, less critical for synthetic data
- id: VERIFICAR-050
github_issue: null
item_type: task
title: Implement large-scale data generation pipeline
status: completed
priority: high
assigned_to: null
created: 1970-01-01T00:00:00Z
updated: 2025-11-26T12:52:18.988989228+00:00
spec: null
acceptance_criteria:
- Generate 10K+ verified (source, target, correctness) tuples
- Parallel generation with rayon (multicore utilization)
- Progress tracking and estimated time remaining
- Automatic Parquet sharding (1GB chunks)
- Support all sampling strategies (exhaustive, coverage-guided, swarm, boundary)
phases: []
subtasks: []
estimated_effort: 8-10 hours
labels:
- pipeline
- performance
- data
notes: Critical path for ML training. BLOCKER - alimentar v0.1.0 is scaffold only, need data loading infrastructure first.
- id: VERIFICAR-051
github_issue: null
item_type: task
title: Implement bug prediction model training
status: completed
priority: high
assigned_to: null
created: 1970-01-01T00:00:00Z
updated: 2025-11-26T12:47:52.431809965+00:00
spec: null
acceptance_criteria:
- Train RandomForest on generated data (10K+ examples)
- 80/20 train/test split with stratification
- Track precision, recall, F1, AUC metrics
- Save trained model to disk (bincode/serde)
- Cross-validation (5-fold)
phases: []
subtasks: []
estimated_effort: 6-8 hours
labels:
- ml
- training
- aprender
notes: Use aprender RandomForestClassifier trained on CodeFeatures -> bug probability
- id: VERIFICAR-052
github_issue: null
item_type: task
title: Implement model evaluation and benchmarking
status: completed
priority: high
assigned_to: null
created: 1970-01-01T00:00:00Z
updated: 2025-11-26T12:49:55.397663221+00:00
spec: null
acceptance_criteria:
- Confusion matrix visualization
- ROC curve and AUC calculation
- Feature importance analysis
- Benchmark inference speed (predictions/sec)
- Comparison baseline vs trained model
phases: []
subtasks: []
estimated_effort: 6-8 hours
labels:
- ml
- evaluation
- metrics
notes: Critical for understanding model quality and production readiness
- id: VERIFICAR-053
github_issue: null
item_type: task
title: Implement RL test prioritizer training loop
status: completed
priority: medium
assigned_to: null
created: 1970-01-01T00:00:00Z
updated: 2025-11-26T12:51:00.270106586+00:00
spec: null
acceptance_criteria:
- Online training with Thompson Sampling
- Track regret over time (cumulative bug detection)
- Save/load prioritizer state (Beta distribution parameters)
- Adaptive exploration rate decay
- Comparison vs random baseline
phases: []
subtasks: []
estimated_effort: 4-6 hours
labels:
- ml
- rl
- prioritization
notes: Build on VERIFICAR-021 with actual training loop
- id: VERIFICAR-060
github_issue: null
item_type: task
title: Integrate with decy (C-to-Rust) transpiler
status: completed
priority: high
assigned_to: null
created: 1970-01-01T00:00:00Z
updated: 2025-11-26T11:41:47.836919826+00:00
spec: null
acceptance_criteria:
- DecyTranspiler trait implementation
- Validate against CPython subset (100+ functions)
- Oracle verification with gcc + rustc
- Bug detection on real C code
phases: []
subtasks: []
estimated_effort: 10-12 hours
labels:
- transpiler
- decy
- integration
notes: High value - enables CPython, Git, SQLite verification
- id: VERIFICAR-061
github_issue: null
item_type: task
title: Integrate with bashrs transpiler
status: completed
priority: medium
assigned_to: null
created: 1970-01-01T00:00:00Z
updated: 2025-11-26T11:36:59.378359752+00:00
spec: null
acceptance_criteria:
- BashrsTranspiler trait implementation
- Validate against bash scripts corpus (500+ scripts)
- Oracle verification with bash + compiled Rust
- Automated dogfooding (bashrs validates itself)
phases: []
subtasks: []
estimated_effort: 6-8 hours
labels:
- transpiler
- bashrs
- integration
notes: Grammar already implemented (VERIFICAR-013), need full integration
- id: VERIFICAR-062
github_issue: null
item_type: task
title: Validate depyler with advanced pattern corpus
status: completed
priority: high
assigned_to: null
created: 1970-01-01T00:00:00Z
updated: 2025-11-26T11:11:31.093559254+00:00
spec: null
acceptance_criteria:
- Run AdvancedDepylerPatternGenerator against depyler v3.20.0
- Track compilation success rate (target >90%)
- Identify remaining type inference gaps
- Generate regression test corpus
phases: []
subtasks: []
estimated_effort: 4-6 hours
labels:
- transpiler
- depyler
- validation
notes: Leverages DEPYLER-0523→0527 fixes. AdvancedDepylerPatternGenerator ready in v0.3.2.
- id: VERIFICAR-070
github_issue: null
item_type: task
title: Optimize generation performance with parallel execution
status: completed
priority: medium
assigned_to: null
created: 1970-01-01T00:00:00Z
updated: 2025-11-26T11:33:53.380629479+00:00
spec: null
acceptance_criteria:
- Parallel generation with rayon (utilize all cores)
- 10x speedup on multicore systems
- Memory-efficient streaming (no OOM on 100K+ examples)
- Benchmark generation throughput (examples/sec)
phases: []
subtasks: []
estimated_effort: 6-8 hours
labels:
- performance
- parallel
notes: Critical for large-scale data generation
- id: VERIFICAR-071
github_issue: null
item_type: task
title: Implement incremental data generation and caching
status: completed
priority: low
assigned_to: null
created: 1970-01-01T00:00:00Z
updated: 2025-11-26T11:52:08.558450186+00:00
spec: null
acceptance_criteria:
- Cache verified examples (avoid re-verification)
- Incremental generation (resume from checkpoint)
- Deduplication (avoid redundant examples)
- SQLite index for fast lookup
phases: []
subtasks: []
estimated_effort: 4-6 hours
labels:
- performance
- caching
notes: Nice-to-have for iterative development
- id: VERIFICAR-080
github_issue: null
item_type: task
title: Create CLI for end-to-end pipeline
status: completed
priority: high
assigned_to: null
created: 1970-01-01T00:00:00Z
updated: 2025-11-26T11:27:46.161925313+00:00
spec: null
acceptance_criteria:
- verificar generate --count 10000 --strategy swarm --output data/
- verificar verify --input data/ --transpilers depyler,bashrs,decy
- verificar train --input data/verified/ --output models/
- verificar evaluate --model models/bug_predictor.bin --test data/test/
- Rich progress bars (indicatif)
phases: []
subtasks: []
estimated_effort: 8-10 hours
labels:
- cli
- ux
notes: User-facing interface for entire pipeline
- id: VERIFICAR-081
github_issue: null
item_type: task
title: Package and publish to crates.io
status: completed
priority: medium
assigned_to: null
created: 1970-01-01T00:00:00Z
updated: 1970-01-01T00:00:00Z
spec: null
acceptance_criteria:
- Published to crates.io as verificar v0.3.2 ✅
- Complete README with quickstart ✅
- API documentation (docs.rs) ✅
- CHANGELOG with version history ✅
phases: []
subtasks: []
estimated_effort: 4-6 hours
labels:
- release
- documentation
notes: v0.3.2 published 2025-11-25 with advanced depyler patterns
- id: VERIFICAR-090
github_issue: null
item_type: task
title: Implement LLM fine-tuning integration with entrenar
status: completed
priority: low
assigned_to: null
created: 1970-01-01T00:00:00Z
updated: 1970-01-01T00:00:00Z
spec: null
acceptance_criteria:
- Export verified tuples to JSON for LLM training
- Integration with entrenar for LoRA fine-tuning
- Prompt templates for code-to-code translation
- Evaluation on held-out test set
phases: []
subtasks: []
estimated_effort: 12-16 hours
labels:
- ml
- llm
- entrenar
notes: Advanced - requires entrenar stability
- id: VERIFICAR-091
github_issue: null
item_type: task
title: Implement semantic equivalence oracle (beyond I/O)
status: completed
priority: low
assigned_to: null
created: 1970-01-01T00:00:00Z
updated: 2025-11-26T13:18:00.114430775+00:00
spec: null
acceptance_criteria:
- AST-based semantic similarity
- Memory layout equivalence
- Performance profile matching
- Formal verification integration (bounded model checking)
phases: []
subtasks: []
estimated_effort: 16-20 hours
labels:
- oracle
- formal-verification
notes: Research-grade feature - I/O oracle sufficient for v0.4
- id: GH-2
github_issue: null
item_type: task
title: 'New task: GH-2'
status: completed
priority: medium
assigned_to: null
created: 2025-11-26T12:30:36.490288258+00:00
updated: 2025-11-26T12:33:31.972128619+00:00
spec: null
acceptance_criteria: []
phases: []
subtasks: []
estimated_effort: null
labels: []
notes: null