Aider vs Goose (2026)

Aider works through git, lint and test commands, suggested shell commands, web scraping and image and voice input, with no MCP client or plugin system.

Architect mode hands a change from a planning model to a separate editor model, and aider then lints, tests and tries to fix errors when a command fails.

Recipes are shareable packaged workflows with a documented YAML or JSON schema, sub-recipes compose them into larger workflows, subagents run them in isolated contexts and concurrently, and a scheduler runs them unattended.

Work starts from the person through the terminal chat, a one shot --message flag or watch mode comments, with nothing on a schedule, webhook or repository event.

A scheduler runs recipes on cron or a set frequency without a person starting them, on top of the desktop app, CLI, REST API and a gateway for remote access.

A repository map ranks the repo's files and key symbols over a dependency graph to fit a token budget, with read only files and scraped pages as extra context.

Grounding comes from direct file and command access and from MCP extensions that reach outside content, with project guidance read from AGENTS.md, and no codebase indexing, embedding or retrieval layer is documented.

Chat history is saved to a file and can be reloaded on the next launch, with no stated lifetime or purge path.

The built in Memory extension keeps what the user teaches goose as files per project or globally and loads them into every session, but the agent retains nothing on its own.

Manual Approval asks before any tool or extension use, Smart Approval flags riskier actions for approval and Chat Only allows no tool use, though Completely Autonomous is the default.

Aider is a local CLI with no hosted service, accounts or organization layer, so no SSO, roles, audit logging or attestation is offered.

The agent runs locally under the user's own credentials with tools that can be disabled and scoped per recipe and subagent, and with no hosted service, accounts or organization layer it offers no SSO, RBAC, audit logging or attestation.

Every AI edit is a git commit with a generated message that /diff shows and /undo reverts, and the session and full LLM conversation can be logged.

Every session is kept in a local SQLite database with the full conversation of messages and tool interactions, browsable and resumable in the CLI and desktop app, though with no organization level export.

A published Recipe Cookbook offers ready to use recipes shareable by URL and launchable in one click, alongside more than 70 installable MCP extensions and a skills system for role based specialization.

Aider can be scripted with --message or its Python Coder class, but the Python API is unsupported and no HTTP API, SDK or MCP server is offered.

The repository ships a benchmark harness that measures pass rates, cost and timing for aider with a chosen model, and the project publishes its polyglot leaderboard from it.

The repository's evals/harbor directory runs goose against benchmark tasks so a customer can evaluate the agent's behavior with the project's own harness, and recipes add structured response schemas and retries for deterministic checks.

Aider edits files, runs shell commands and scrapes web pages into markdown, but it does not drive a browser or operate software through a graphical interface.

Computer control comes from the built in Computer Controller extension, which drives macOS apps through the third party Peekaboo CLI and scrapes the web, while browser control comes from catalog extensions such as Playwright, Selenium and Browserbase that the user installs.