act101

act101 is a groundbreaking developer tool that fundamentally changes how AI agents interact with source code. Unlike conventional AI coding assistants that rely on whole-file rewrites or text-based patches, act101 provides a Model Context Protocol (MCP) server that exposes 183 AST-aware refactor operations, 30 codebase analyzers, 15 query operations, and 8 porting operations to AI coding agents. Built as a single native Rust binary with zero external dependencies, act101 supports 163 programming language grammars, making it the most comprehensive language-aware code manipulation tool available. The product is designed for developers using AI coding agents like Claude Code, Cursor, Codex, and OpenCode who need precise, undoable, and context-aware code transformations. Its core value proposition is enabling AI agents to perform actual refactoring and code porting tasks rather than simply generating text that looks like code. All operations execute locally with no telemetry, no indexing, and no caching, ensuring your code never leaves your machine. The tool is free for personal use and includes automatic checkpointing with instant undo on every operation, giving developers complete confidence when delegating complex code transformations to AI agents.

AST-Aware Refactor Operations

act101 provides 183 distinct AST-aware refactor operations that go far beyond simple find-and-replace or whole-file rewrites. These operations include extract-function, rename-symbol, move-symbol, inline-variable, convert-to-dataclass, extract-trait, add-type-hints, organize-imports, and 174 more. Each operation understands the abstract syntax tree of the target language, ensuring transformations maintain syntactic validity, preserve comments and formatting, and handle cross-file dependencies correctly. When an AI agent calls extract-function on a Python method, act101 automatically identifies all local variables that need to become parameters, handles return values, and updates all call sites across the entire codebase. Every operation creates an automatic checkpoint, allowing instant undo if the result is not satisfactory.

Cross-Language Porting State Machine

The porting system in act101 implements a sophisticated four-phase state machine that drives end-to-end language migrations. The process begins with port_contract, which establishes the source-to-target mapping and migration scope. Next, port_inventory enumerates every symbol, type, and structure that must be moved. The port_order phase resolves dependency ordering to ensure the migration proceeds in a logically correct sequence. Finally, the port_manifest state machine tracks progress through init, add, update, remove, and note states. This structured approach enables AI agents to port code between any two of the 163 supported grammars, such as C to Rust, Ruby to Elixir, or COBOL to Java, with predictable, auditable results.

Comprehensive Codebase Analyzers

act101 includes 30 built-in codebase analyzers that provide AI agents with deep structural understanding of a codebase before any transformations begin. These analyzers cover cohesion, coupling, cycles, chokepoints, hotspots, dead code detection, module boundary analysis, layers, seams, clusters, surface area, fan balance, migration readiness, and type completeness. Each analyzer returns structured data that AI agents can use to make informed decisions about refactoring strategies, identify problematic areas, and prioritize work. The analysis is performed on-demand with no pre-indexing or caching, meaning results are always current and reflect the exact state of the codebase at the time of the query.

Pre-Built Agent Skills

act101 ships with 10 pre-built agent skills that compose individual operations into complete engineering workflows. These skills include architecture-audit, code-review, refactoring, code-navigation, code-generation, migration-assessment, boundary-analysis, change-impact, health-check, and architectural-refactoring. Each skill orchestrates multiple operations and analyzers to accomplish a higher-level goal. For example, invoking the architecture-audit skill in Claude Code triggers a sequence of coupling analysis, cycle detection, and boundary analysis, then presents a comprehensive architectural report. Skills are invoked using a simple /skill-name command within any MCP-aware client, making complex workflows accessible with a single command.

Large-Scale Code Refactoring with AI Agents

Development teams can use act101 to delegate complex refactoring tasks to AI agents with confidence. When a team needs to extract a reusable library from a monolithic codebase, the AI agent can systematically apply extract-function, move-symbol, and organize-imports operations across hundreds of files. The AST-aware nature of these operations ensures that all references, imports, and dependencies are correctly updated. Automatic checkpointing on every operation means the team can review each change and instantly roll back any transformation that does not meet expectations. This approach reduces the manual effort of large-scale refactoring from weeks to hours while maintaining code quality and consistency.

Cross-Language Code Migration Projects

Organizations modernizing legacy systems can leverage act101 for automated language migration. For example, migrating a legacy COBOL banking application to Java involves understanding complex business logic, data structures, and control flow. The AI agent uses port_contract to define the migration scope, port_inventory to catalog every COBOL data structure and procedure, and the port_manifest state machine to track progress. The agent can then systematically translate each component while maintaining the original business logic. The structured state machine approach ensures no symbols are missed and dependencies are resolved in the correct order, resulting in a complete, compilable Java codebase.

Code Quality Auditing and Improvement

Engineering teams can run comprehensive code quality audits using act101's analyzer suite. Before a major release, a team might invoke the architecture-audit skill to identify coupling issues, circular dependencies, and dead code across their entire codebase. The AI agent presents a structured report with specific locations and recommendations. The team can then immediately begin addressing the identified issues using the refactoring skills, with the agent making targeted, undoable changes. This workflow transforms code quality improvement from a manual, error-prone process into a systematic, AI-assisted activity that produces measurable results.

Automated Code Review and Change Impact Analysis

When developers submit pull requests, act101 can perform automated change impact analysis using its 30 analyzers and 15 query operations. The AI agent analyzes the proposed changes, identifies all affected code paths, detects potential regressions, and checks for boundary violations. The agent can then generate a detailed code review report that includes specific recommendations for improvement. For complex changes, the agent can also suggest alternative refactoring strategies that would reduce coupling or improve testability. This use case helps teams maintain code quality standards while accelerating the review process.

How does act101 ensure my code remains private and secure?

act101 operates entirely on your local machine as a single native Rust binary. There is no telemetry, no data collection, and no code transmission to external servers. The only network communication is for license verification, which occurs infrequently and does not involve your source code. All parsing, analysis, and transformation operations execute locally with zero external dependencies. This architecture eliminates supply chain attack surface and ensures your intellectual property never leaves your control.

What programming languages does act101 support?

act101 supports 163 programming language grammars in a single binary. This includes all major languages such as Python, JavaScript, TypeScript, Rust, Go, Java, C, C++, Ruby, Elixir, COBOL, and hundreds more. The language support covers both modern and legacy languages, making act101 suitable for migration projects involving older codebases. The grammars are compiled into the binary at build time, so there is no need to download or configure language-specific plugins.

How does act101 compare to using AI agents with whole-file rewrites?

Traditional AI agents use whole-file rewrites that often lose comments, break formatting, and provide no undo capability. act101 provides typed, AST-aware operations that preserve all formatting and comments, maintain cross-file consistency, and create automatic checkpoints for instant undo. Additionally, act101 operations use approximately 85 percent fewer tokens than equivalent file-based operations, making them more efficient and cost-effective. The structured operations also produce more predictable results that are easier to review and validate.

Can act101 be used with any AI coding agent?

Yes, act101 implements the Model Context Protocol (MCP) standard, making it compatible with any MCP-aware client. This includes Claude Code, Cursor, Windsurf, Codex, OpenCode, and other AI coding tools that support MCP. The built-in MCP server exposes all operations, analyzers, and skills through a standardized interface, so you can use the same powerful capabilities regardless of which AI agent you prefer. Configuration is straightforward and documented in the installation guide.

act101 is free for personal use with no limitations on features or operations. For commercial use and team deployments, please see the pricing page on the act101 website. The free tier includes access to all 163 grammars, all 183 refactor operations, all 30 codebase analyzers, all 15 query operations, all 8 porting operations, and all 10 pre-built agent skills. There are no token limits, no usage caps, and no feature restrictions for personal users.