Project Roadmap

Current Version: 5.8.1 | Release Date: 2026-04-05 | Status: Active Development

Executive Summary

ContribAI is a mature autonomous AI contribution system. Originally built in Python (v0.x–v4.0), it was rewritten in Rust (v5.0.0) for performance, safety, and new capabilities like tree-sitter AST parsing and PageRank file ranking. v5.2.0 added an interactive TUI and full CLI parity. v5.3.0 introduced watchlist mode and 13-language AST. v5.4.0 added dream memory consolidation and risk classification. v5.5.0 delivers multi-file PRs, end-to-end issue solving, and conversation memory. v5.6.0 adds integration tests, LLM retry with backoff, GitHub rate limiter, and the doctor command. v5.7.0–v5.8.0 deliver cross-file import resolution, hunt CLI fix, and full integration test coverage for hunt/patrol. v6.2.0 adds closed-PR failure analysis and outcome-aware quality scoring. The roadmap focuses on enterprise scalability and plugin ecosystem.

Release Timeline

v0.x Series (2026-03-17 to 2026-03-20) — Foundation Building (Python)

Version	Date	Milestone	Status
v0.1	2026-03-17	Core pipeline (discovery → analysis → generation → PR)	✓ Complete
v0.4	2026-03-18	Web dashboard + REST API	✓ Complete
v0.5	2026-03-18	Scheduler + cron automation	✓ Complete
v0.7	2026-03-19	Multi-LLM support (Gemini, OpenAI, Anthropic, Ollama)	✓ Complete
v0.11	2026-03-20	Hunt Mode (autonomous multi-round hunting)	✓ Complete

v1.x Series (2026-03-20) — Quality & Safety (Python)

Version	Date	Milestone	Status
v1.0	2026-03-20	Official release; quality improvements	✓ Complete
v1.5	2026-03-20	CLA/DCO handling; compliance automation	✓ Complete
v1.8	2026-03-20	Cross-file pattern detection + bulk fixes	✓ Complete

v2.x Series (2026-03-22 to 2026-03-26) — Learning & Resilience (Python)

Version	Date	Milestone	Status
v2.0	2026-03-22	Safety gates (quality scorer, duplicate prevention)	✓ Complete
v2.2	2026-03-23	PR Patrol (auto-monitor + auto-fix feedback)	✓ Complete
v2.4	2026-03-25	Outcome memory (learns from PR results)	✓ Complete
v2.6	2026-03-26	MCP server (14 tools for Claude Desktop)	✓ Complete
v2.7	2026-03-26	Event bus (15 typed events) + working memory	✓ Complete
v2.8	2026-03-26	Context compression + progressive skills	✓ Complete

v3.x–v4.x Series (2026-03-26 to 2026-03-28) — Production Hardening (Python)

Version	Date	Milestone	Status
v3.0.0	2026-03-26	EventBus system, Formatter, MCP Client, CLI flags	✓ Complete
v3.0.4	2026-03-28	Security hardening (constant-time API keys, webhook validation)	✓ Complete
v3.0.6	2026-03-28	SKIP_DIRECTORIES filter, auto-close linked issues	✓ Complete
v4.0.0	2026-03-28	Repo Intelligence, Smart Dedup, Issue-First Strategy	✓ Complete

v5.0.0 (2026-03-28 to 2026-03-31) — Rust Rewrite ✓

Sprint	Date	Milestone	Status
S1	2026-03-28	Core architecture: config, models, middleware, events, errors	✓ Complete
S2	2026-03-29	CI Monitor, Outcome Learning, Vertex AI, token cache	✓ Complete
S3	2026-03-30	Web dashboard (axum), web-server CLI command	✓ Complete
S4	2026-03-30	GraphQL GitHub, JSONL event log, template+plugin system	✓ Complete
Parity	2026-03-31	Full Python→Rust feature parity audit + gap implementation	✓ Complete

Key Achievements (v5.0.0):

✓ Complete Python→Rust rewrite: 62 .rs files, ~21,400 LOC, 323 tests
✓ Tokio async runtime (replacing Python asyncio)
✓ Axum web framework (replacing Python FastAPI)
✓ Clap CLI with 21 commands (replacing Python Click)
✓ rusqlite for SQLite (replacing Python aiosqlite)
✓ serde for serialization (replacing Python Pydantic)
✓ MCP server expanded: 21 tools (was 14)
✓ API key auth with constant-time comparison
✓ HMAC-SHA256 webhook verification
✓ 17 analysis skills (5 new vs Python)

v5.2.0 (2026-04-01) — Interactive Login, Rust CI, 4-platform Binaries ✓

Key Achievements (v5.2.0):

✓ Interactive contribai login — switch LLM providers, update tokens, launch wizard
✓ One-line install scripts (install.sh + install.ps1) — auto-detect OS/arch
✓ 4-platform release binaries: Linux x86_64, Windows x86_64, macOS Intel, macOS ARM64
✓ Rust-first CI pipeline: fmt + clippy -D warnings + 335 tests + cargo audit
✓ Python legacy tests only trigger with [python] commit label
✓ Resolved 24+ clippy warnings, zero-warning strict lint
✓ Cross-compilation support for macOS aarch64

v5.1.0 (2026-04-01) — Interactive TUI & Full CLI Parity ✓

Key Achievements (v5.1.0):

✓ Interactive TUI: ratatui 4-tab browser (Dashboard/PRs/Repos/Actions)
✓ Real notify-test: live HTTP to Slack, Discord, Telegram
✓ Full 22-command CLI (init, login, leaderboard, models, templates, profile, config-get/set/list, system-status, notify-test)
✓ Setup wizard (contribai init)
✓ Config editor (config-get, config-set, config-list)
✓ 63 .rs files, ~22,000 LOC, 335 tests
✓ Python moved to python/ (v4.1.0 legacy, preserved for reference)
✓ Root Cargo.toml workspace — cargo build from project root

v5.3.0 (2026-04-02) — Watchlist Mode & 13-Language AST ✓

Key Achievements (v5.3.0):

✓ Watchlist mode: targeted repo scanning for focused ecosystem work
✓ Rotating sort order + pagination for diverse discovery across hunt rounds
✓ Expanded AST support: 13 languages (was 8) including Ruby, PHP, Bash, YAML, JSON
✓ All-language discovery (scan all 15 supported languages by default)
✓ Gemini 3.x model support

v5.4.0 (2026-04-03) — Dream Memory & Risk Classification ✓

Key Achievements (v5.4.0):

✓ Dream memory consolidation: efficient memory entry consolidation during idle periods
✓ Risk classification: Low/Medium/High risk levels for auto-submit control
✓ Conversation-aware patrol: maintains context history for intelligent feedback responses
✓ Enhanced PR lifecycle management

v5.4.2 (2026-04-04) — Bug Fixes & Polish ✓

Key Achievements (v5.4.2):

✓ Auto-clean 404 PRs from patrol monitoring
✓ Config-set YAML list values: proper quoting for list items
✓ MCP stdout fix: tracing + banner redirected to stderr
✓ 65 .rs files, ~26,000+ LOC, 355 tests
✓ 40+ CLI commands (expanded from 22)

v5.5.0 (2026-04-04) — Multi-File PRs, Issue Solver, Conversation Memory ✓

Key Achievements (v5.5.0):

✓ Multi-file PR batching: pipeline merges related findings into single multi-file PR
✓ Issue solver end-to-end: solve command generates code + creates PRs with Fixes #N linking
✓ PR conversation memory: patrol stores full threads in SQLite, injects history for context-aware LLM responses
✓ Dream profile wiring: pipeline filters rejected contribution types using repo outcome history
✓ Auto-dream trigger on run_targeted() path
✓ 66 .rs files, ~28,000 LOC, 355 tests

v5.6.0 (2026-04-04) — Integration Tests, Merge Rate, LLM Retry ✓

Key Achievements (v5.6.0):

✓ Integration test framework: wiremock 0.6 + MockLlm test infrastructure
✓ 33+ new tests (388 total): pipeline, patrol, hunt pre-processing paths
✓ LLM retry with exponential backoff: configurable retries for transient failures
✓ GitHub rate limiter: token-bucket rate limiting for API calls
✓ doctor command: system health diagnostics (config, LLM, GitHub, DB)
✓ DB indexes for hot query paths (analyzed_repos, submitted_prs, pr_outcomes)
✓ Semantic code chunking: intelligent truncation respecting AST boundaries

v5.7.0 (2026-04-05) — Hunt CLI Fix, TTL, Semantic Chunking ✓

Key Achievements (v5.7.0):

✓ Hunt CLI fix: hunt command now calls pipeline.hunt() instead of pipeline.run()
✓ cargo fmt --all: full codebase formatting pass

v5.8.0 (2026-04-05) — Cross-File Import Resolution, Integration Tests ✓

Key Achievements (v5.8.0):

✓ Cross-file import resolution: 1-hop resolution for 5 languages (Rust/Python/JS-TS/Go/Java), capped at 20 symbols
✓ symbol_map wired in pipeline: type context feature now has real data from extract_symbols()
✓ GitHubClient with_base_url(): wiremock-friendly testability without prod code changes
✓ Mock GitHub infrastructure: mock_github.rs with fixture factories + wiremock helpers
✓ Patrol integration tests (5 tests): bot filtering, classification, conversation context, 404 auto-clean, dry-run
✓ Hunt integration tests (4 tests): daily limit gate, merge-friendly filter, TTL skip, empty discovery
✓ 67 .rs files, ~29,200 LOC, 413 tests

v6.2.0 (2026-04-05) — Closed-PR Analysis, Outcome-Aware Scoring ✓

Key Achievements (v6.2.0):

✓ Closed-PR failure analysis: patrol fetches review comments + CI status for closed-but-not-merged PRs, stores feedback via memory.record_outcome()
✓ Outcome-aware quality scoring: 8th check check_outcome_history adjusts score based on merge_rate and rejected_types
✓ resolved_imports field on RepoContext: clean cross-file data separated from symbol_map
✓ Depth guard on walk_import_nodes (capped at 8)
✓ Pipeline integration test for process_repo symbol_map wiring with wiremock
✓ 67 .rs files, ~29,577 LOC, 602 tests

Feature Status Matrix (v6.2.0)

Core Pipeline

Feature	Status	Details
Repository discovery	✓ Complete	GitHub Search API (REST + GraphQL)
Multi-strategy analysis	✓ Complete	7 analyzers, 17 skills, framework detection
Tree-sitter AST parsing	✓ Complete	13 language grammars (Rust-only, v5.3.0)
PageRank file ranking	✓ Complete	Import graph analysis (Rust-only)
12-signal triage	✓ Complete	Multi-factor issue scoring (Rust-only)
LLM-powered generation	✓ Complete	Multi-provider routing, self-review, quality scoring
Risk classification	✓ Complete	Low/Medium/High for auto-submit control (v5.4.0)
Autonomous PR creation	✓ Complete	Fork, branch, commit, PR, CLA/DCO handling
Hunt mode (multi-round)	✓ Complete	Configurable rounds, delays, deduplication, watchlist (v5.3.0)
Dream memory	✓ Complete	Consolidate memory during idle periods (v5.4.0)
Cross-file fixes	✓ Complete	Bulk fix for pattern repetition
Issue-driven solving	✓ Complete	Fetch + solve open GitHub issues

Safety & Compliance

Feature	Status	Details
Rate limiting	✓ Complete	Daily PR limit + API rate respect
Quality gate	✓ Complete	7-check scorer, 0.6 min threshold
Duplicate prevention	✓ Complete	Fuzzy title matching (>90% = duplicate)
AI policy detection	✓ Complete	Parse CONTRIBUTING.md for AI bans
CLA auto-signing	✓ Complete	CLA-Assistant, EasyCLA
DCO signoff	✓ Complete	Auto-append to all commits
Deep validation	✓ Complete	LLM validates findings vs. file context
Webhook verification	✓ Complete	HMAC-SHA256 signature validation
API key auth	✓ Complete	Constant-time comparison (timing attack safe)
Auto-clean 404 PRs	✓ Complete	Remove stale forks with 404 status (v5.4.2)

Platform Features

Feature	Status	Details
Web dashboard	✓ Complete	Axum REST API + static UI at `:8787`
Scheduler	✓ Complete	Tokio-based cron automation
Webhooks	✓ Complete	GitHub webhook receiver with HMAC verification
Profiles	✓ Complete	Named presets (security-focused, docs-focused, etc.)
Templates	✓ Complete	Built-in contribution templates
Plugins	✓ Complete	Trait-based plugin system
Notifications	✓ Complete	Slack, Discord, Telegram
MCP server	✓ Complete	21 tools for Claude Desktop
MCP client	✓ Complete	StdioMcpClient for external MCP servers

Architecture & Internals

Feature	Status	Details
Event bus	✓ Complete	18 typed events, JSONL file logging
Sub-agent registry	✓ Complete	5 agents (Analyzer, Generator, Patrol, Compliance, Issue)
Context compression	✓ Complete	3-tier with language-aware signature extraction
Working memory	✓ Complete	Per-repo cache with 72h TTL
Outcome learning	✓ Complete	PR outcome tracking + repo preferences
Error handling	✓ Complete	thiserror enum hierarchy + graceful recovery
Async-first design	✓ Complete	All I/O via Tokio async

Completed Milestones

Milestone 1: MVP (v0.1–v0.11, Python) ✓

✓ Pipeline: discovery → analysis → generation → PR
✓ 7 multi-strategy analyzers, multi-LLM, hunt mode, web dashboard

Milestone 2: Safety & Learning (v1.0–v2.8, Python) ✓

✓ Quality scoring, CLA/DCO, PR patrol, outcome memory, event bus

Milestone 3: Production Hardening (v3.0–v4.0, Python) ✓

✓ MCP server, enhanced code gen, security hardening, comprehensive docs

Milestone 4: Rust Rewrite (v5.0.0) ✓

✓ Full feature parity with Python (99%+)
✓ New Rust-only capabilities (tree-sitter, PageRank, triage, compression)
✓ 62 files, ~21,400 LOC, 323 tests, single static binary

Milestone 5: Full CLI Parity + TUI (v5.2.0) ✓

✓ 22 CLI commands (was 21)
✓ Interactive ratatui TUI browser
✓ Real notification testing (Slack, Discord, Telegram)
✓ Setup wizard + config editor commands
✓ Python moved to python/ legacy, root Cargo workspace
✓ 335 tests

Milestone 6: Advanced Features (v5.3.0–v5.5.0) ✓

✓ Watchlist mode for targeted repo scanning
✓ 13-language AST support (expanded from 8)
✓ Dream memory consolidation for efficiency
✓ Risk classification for intelligent auto-submit
✓ Conversation-aware PR patrol with context
✓ Auto-clean 404 PRs from patrol
✓ Multi-file PR batching
✓ Issue solver end-to-end with Fixes #N linking
✓ PR conversation memory for context-aware responses
✓ 40+ CLI commands, 355 tests, 66 .rs files, ~28,000 LOC

Milestone 7: Test Coverage & Analysis (v5.6.0–v6.2.0) ✓

✓ Integration test framework (wiremock 0.6 + MockLlm)
✓ Hunt & patrol integration tests (9 tests covering critical paths)
✓ Cross-file import resolution (5 languages, 1-hop, 20-symbol cap)
✓ symbol_map wired for type-aware code generation
✓ LLM retry with exponential backoff
✓ GitHub rate limiter + doctor command
✓ DB indexes for hot query paths
✓ Closed-PR failure analysis + outcome-aware quality scoring (8-check)
✓ 67 .rs files, ~29,577 LOC, 602 tests

v6.1.0 (2026-04-11) — Code Quality & Framework Detection ✓

Key Achievements (v6.1.0):

✓ 0 clippy warnings (6 → 0) — strict lint enforced
✓ Framework detection from imports (20+ frameworks: Django, React, Rails, etc.)
✓ Copilot provider fully wired in all factory functions
✓ Session dead code removed
✓ 602 tests (+15 framework detection tests)

v6.2.0 (2026-04-11) — Dependencies, Benchmarks & Optimization ✓