A catalog of outside tools for AI agents and developers, one file per tool, with notes from real use.
Warning
The toolbox is temporary. A searchable tool library replaces it completely after Flow's full release.
Flow downloads it to ~/.flow/repos/toolbox/ when it installs, and /flow:research searches it before the web. Without Flow, clone it:
$ git clone https://github.com/Adrian333Dev/toolbox.git- Pick every folder in the tree below that could hold the answer. Often that is several.
- Print the one-line description of every tool in those folders, in one search:
grep -r "^description:" --exclude="*.readme.md" agent-tools/memory agent-tools/code. Each line comes back with its file's path. - Open the 2 or 3 files that look right and read their notes. A collection's file also lists its skills, plugins and subagents.
When no folder fits, search every file for a word, notes and saved READMEs included: grep -ril "lip-sync" agent-tools software.
agent-tools/ holds tools made for AI agents, software/ holds everything else, and inside each a folder groups tools by what they help with. Contributing covers adding a tool, and rebuilding this tree.
.
├── agent-tools/ // Made for an AI agent to use or run inside, even partly
│ ├── behavior/ // Changing how the agent answers or thinks
│ ├── browser/ // Driving a browser, and pulling content out of web pages
│ ├── code/ // Reviewing, understanding and finding your way around code
│ ├── collections/ // Sets and indexes of skills, plugins, subagents and prompts
│ ├── extending/ // Making skills, plugins and hooks
│ ├── harnesses/ // Programs agents run inside, and toolkits for building your own agent
│ ├── marketing/ // SEO, ads and written content
│ ├── memory/ // Memory that lasts between sessions
│ ├── models/ // Routing requests across model providers
│ ├── research/ // Researching a topic and writing up what was found
│ ├── security/ // Scanning code, reverse engineering and authorized pentests
│ ├── services/ // Connecting the agent to outside services: databases, payments, errors, hosting, automation
│ ├── ui/ // Designing UI, and design systems an agent can copy
│ ├── video/ // Making, cutting and watching video
│ └── workflows/ // Whole development methods to adopt
├── inbox/ // New tools from bin/tool.js add, waiting to be filed
└── software/ // Everything else: libraries, frameworks, services and apps
├── backend/ // Databases, payments and logins
├── documents/ // Converting documents
├── games/ // Game engines, and games built with agents
├── models/ // Testing and changing language models
├── scraping/ // Scraping websites
├── security/ // Security references and access rules
├── ui/ // Components, icons, editors, canvas and maps
├── video/ // Rendering video from code, and templates to start from
└── voice/ // Speech to text and text to speech