A composable agent runtime — pair any frontend with any agent backend over one shared extension layer
Shared types, permissions, ledger, cache, and LLM tool-adapter emitters for @agent-sh/harness-* tools
Umbrella package and PI extension for @agent-sh/harness-* agent tools (read, write, grep, glob, bash, webfetch, websearch, lsp, skill)
Linux desktop-control MCP server: AT-SPI accessibility trees, Wayland/X11 input, screenshots, and compositor window targeting.
Glob tool for AI agent harnesses — ripgrep-backed filename pattern search with structured output, pagination, .gitignore respect, and mtime sort
Grep tool for AI agent harnesses — ripgrep-backed content search with structured output modes, pagination, .gitignore respect, and a WASM-bundled engine
[](https://www.npmjs.com/package/@aws-sdk/middleware-user-agent) [](https://www.npmjs.c
Write / Edit / MultiEdit tools for AI agent harnesses — ledger-anchored, atomic, with structured fuzzy diagnostics
Bash tool for AI agent harnesses — autonomous shell execution with pluggable executor, permission hook, cwd-carry, inactivity timeout, and background job support
Read tool for AI agent harnesses — safe, bounded, line-numbered file reading with pluggable I/O, permissions, ledger, and cache
[](https://www.npmjs.com/package/@aws-sdk/util-user-agent-browser) [](https://www.n
WebFetch tool for AI agent harnesses — HTTP GET/POST with SSRF defense, readability+markdown extraction, redirect-chain reporting, size caps, and per-session cache
Your AI pair programmer
Skill tool for AI agent harnesses — authored SKILL.md activation with progressive disclosure, permission-gated activation, trust-gated project skills, and pluggable SkillRegistry
Turn a function into an `http.Agent` instance
LSP tool for AI agent harnesses — language-server-protocol operations (hover, definition, references, symbols) with .lsp.json manifest, lazy-spawn policy, and pluggable LspClient adapter
Default user agent string for nodejs http request
Maps proxy protocols to `http.Agent` implementations
Global HTTP/HTTPS proxy configurable using environment variables.
the http/https agent used by the npm cli
An HTTP(s) proxy `http.Agent` implementation for HTTPS
HTTP Agent that keeps socket connections alive between keep-alive requests. Formerly part of mikeal/request, now a standalone module.
[](https://www.npmjs.com/package/@aws-sdk/util-user-agent-node) [](https://www.npmjs.com/
Get a user agent string across all JavaScript Runtime Environments
Linux desktop control over MCP — AT-SPI accessibility tree, multi-compositor window targeting (GNOME, KWin, Hyprland, i3, COSMIC), screencast portal screenshots, and ydotool input synthesis. Wayland-first, X11 best-effort.
Shared types and traits for @agent-sh/harness-* Rust tools: ToolError, PermissionPolicy, ReadOperations
Glob tool for AI agent harnesses — bash-style pattern matching with .gitignore enumeration, alias pushback, NOT_FOUND siblings, and auto-split absolute-path patterns
Read tool for AI agent harnesses — filesystem read with pagination, binary sniff, directory listing, attachment shape, NOT_FOUND fuzzy siblings
Grep tool for AI agent harnesses — ripgrep-library content search with structured output modes, .gitignore respect, alias pushback, and fuzzy NOT_FOUND siblings
WebFetch tool for AI agent harnesses — HTTP GET/POST with SSRF defense, readability+markdown extraction, redirect-chain reporting, size caps, and per-session cache
Bash tool for AI agent harnesses — autonomous shell execution with tokio process management, cwd-carry, inactivity timeout, head+tail spill, background jobs
LSP tool for AI agent harnesses — language-server-protocol operations (hover, definition, references, symbols) with manifest, lazy-spawn policy, pluggable client adapter
Write/Edit/MultiEdit tool for AI agent harnesses — atomic write, read-before-edit ledger, OLD_STRING_NOT_UNIQUE with match locations, OLD_STRING_NOT_FOUND with fuzzy candidates, sequential multi-edit pipeline
WebSearch tool for AI agent harnesses — declarative web search via a session-configured SearXNG backend, with SSRF defense, permission hook, result-count cap, and a discriminated result surface
Skill tool for AI agent harnesses — authored SKILL.md activation with progressive disclosure, permission-gated activation, trust-gated project skills, and pluggable SkillRegistry
Umbrella re-export for harness-* AI agent tools: core, read, write, grep, glob, bash, webfetch, websearch, lsp, skill
No description provided.
No description provided.
No description provided.
No description provided.
No description provided.
No description provided.
No description provided.
No description provided.
No description provided.
No description provided.
No description provided.
No description provided.