Extend the agent with your own tools.

Read and write files and directories on the local filesystem, scoped to a directory you choose.

Fetch and convert web pages to Markdown for the model to read.

A persistent knowledge-graph memory the agent can read and write across a session.

Search repos and code, read and comment on issues and pull requests, and manage GitHub via a personal access token.

Run read-only queries against a Postgres database for schema-aware answers.

Drive a real browser to exercise a flow the agent just implemented.

Post messages and read channel history from inside a chat turn.

Web search the agent can use to ground answers in current information.

Read, search, and manipulate Git repositories — status, diffs, commits, branches, and tags — without leaving the conversation.

Get the current time and date and convert across time zones — no more stale-training-cutoff timestamps.

Dynamic, step-by-step reasoning chains the agent can revise and branch as a problem unfolds.

Pull current, version-accurate documentation for libraries and frameworks straight into context — fewer hallucinated APIs.

Fetch issues, stack traces, and release metadata so the agent can debug against real production errors.

Query a local SQLite database with schema-aware answers — great for fixtures, prototypes, and analysis.

Manage containers, images, volumes, and compose stacks from a chat turn.

Inspect and manage cluster resources — pods, deployments, logs — without memorizing kubectl incantations.

Geocode addresses, search places, and get directions for location-aware features.

Search and update Notion pages and databases, so specs and docs stay in sync with the code.

Read and search an Obsidian vault to ground answers in your own notes and knowledge base.

Pull design frames, styles, and specs so generated UI matches the mockup instead of guessing.