Skip to content

The datasheet for every AI agent

Research report · data-report

On the command line: a census of 143 listings

67 of 143 listings answer yes for “On the command line”. Counted from our own datasheet, with the query that reproduces every figure.

Sep 24, 2026updated Oct 3, 20262 min readSource-linked research

On the command line, across 143 settled listings

  1. Yes67 of 143

  2. Qualified30 of 143

  3. No17 of 143

  4. Does not apply25 of 143

  5. Not established4 of 143

Of the 143 Published listings whose datasheet settles “On the command line”, the largest group is “Yes” at 67 (47%). Measured 2026-10-03.

Denominator: 143 Published listings whose datasheet settles “On the command line”.Method: Counted from the stored datasheet answer for “On the command line” across every Published listing, all of which have a recognised stored answer on that row.Measured .
Cite this chart

Reuse this chart anywhere, with credit. Paste this HTML — it links back to the report the numbers come from.

<a href="https://theagentsindex.com/blog/surface-cli-census/" target="_blank" rel="nofollow noopener"><img src="https://theagentsindex.com/charts/surface-cli-census.png" alt="On the command line, across 143 settled listings — The Agents Index" title="On the command line, across 143 settled listings — The Agents Index" width="640" loading="lazy" style="max-width:100%;height:auto" /></a>

Explore the matching agents

On the command line · Datasheet criterion surface_cli · Report counted .

Current listings as of ; this set may differ from the report's original sample. Includes “Yes” and “Qualified” answers only. Read the conditions on qualified answers before treating them as a match. “No”, “Not applicable” and “Not established” answers are excluded.

View 98 current matching records
  • Ada Qualified

    There is no first-party Ada CLI; terminal access exists only by pointing Codex CLI or Gemini CLI at the MCP server to manage the agent.

    Source retrieved 2026-10-03

  • AgentBridge Yes

    The terminal is the only surface: a globally installed abg/agentbridge command with init, claude, codex, resume, pairs, doctor, budget, logs and kill subcommands.

    Source retrieved 2026-10-03

  • AgentsKit.js Yes

    A first-party terminal surface exists: nine agentskit subcommands for scaffolding, one-shot runs, interactive chat and environment diagnosis, plus an Ink terminal chat UI.

    Source retrieved 2026-09-09

  • Agno Yes

    The `agno` command scaffolds a project, starts and stops it, connects clients and mints service-account tokens.

    Source retrieved 2026-09-02

  • Aident Loadout Yes

    There is a first-party aident CLI that handles install, authentication and updates, and the docs state it exposes the same discovery and execution workflow as MCP.

    Source retrieved 2026-10-03

  • Aider Yes

    The terminal is the primary and default surface: you launch aider with filenames and talk to it at its own prompt in the shell.

    Source retrieved 2026-09-30

  • Aisera Agent Composer (Agent Studio) Qualified

    A terminal surface exists but not for agent building: the documented CLI belongs to the LLM Gateway developer tooling and covers managing applications, deploying models and monitoring usage.

    Source retrieved 2026-09-03

  • Amplemarket Qualified

    There is no first-party command-line tool. Terminal use exists only because Claude Code can attach to the MCP server, so the CLI experience is another vendor's client speaking to Amplemarket.

    Source retrieved 2026-09-10

  • Antigravity CLI Yes

    The terminal is the whole product: a lightweight TUI installed as agy that carries the desktop agent's reasoning and multi-file editing into the shell.

    Source retrieved 2026-09-10

  • API.market Qualified

    There is no first-party CLI tool; terminal use is plain HTTP with cURL, which the vendor lists as a supported client for testing and debugging.

    Source retrieved 2026-10-03

  • Argorant Yes

    A published npm CLI runs counts, searches, exports, campaigns and the inbox from a terminal, and needs no install to try.

    Source retrieved 2026-09-26

  • AudioCodes Live Hub Qualified

    AudioCodes ships no CLI of its own; terminal use is indirect, by pointing the Claude Code command-line agent at the platform's MCP endpoint with API client credentials.

    Source retrieved 2026-09-24

  • Augment Code Yes

    Auggie CLI installs from npm and runs interactively or headless.

    Source retrieved 2026-09-02

  • AutoGen Qualified

    A built-in Console UI streams a run to the terminal and UserProxyAgent can take console input, but there is no standalone agent CLI; the autogenstudio command only launches the web UI.

    Source retrieved 2026-09-02

  • Bland Yes

    A published npm CLI drives calls, pathways, numbers, guard rails, evals and releases from a terminal.

    Source retrieved 2026-09-11

  • Botpress Yes

    The adk CLI scaffolds, runs, deploys and debugs agents, and pricing is identical whether you build through the CLI or the visual builder.

    Source retrieved 2026-09-02

  • Botsify Qualified

    The comparison table claims terminal-UI support, but no CLI is documented in the API docs or on any product page.

  • Browserbase Yes

    Browse CLI fetches, searches, browses and manages sessions, contexts and Functions from the terminal.

    Source retrieved 2026-09-02

  • Cartesia Yes

    A `cartesia` CLI creates, deploys and manages agents, and can chat with agent logic locally.

    Source retrieved 2026-09-02

  • Claude Agent SDK Yes

    claude -p is documented as the SDK's CLI surface, with JSON and stream-JSON output for scripted callers.

    Source retrieved 2026-09-02

  • Claude Code Yes

    The terminal CLI is the primary surface, installed by shell script, Homebrew, WinGet or Linux package managers.

    Source retrieved 2026-09-02

  • Cline Yes

    A terminal client installed from npm gives both an interactive session and a headless mode triggered by piping, redirection or the JSON flag.

    Source retrieved 2026-09-12

  • CodeRabbit Yes

    A terminal tool reviews local Git changes before commit and emits structured JSON for coding agents, though the CLI carries an Open Beta badge.

    Source retrieved 2026-09-15

  • Codex CLI Yes

    The terminal is the product's primary surface, installed via installer, npm or Homebrew.

    Source retrieved 2026-09-02

  • CrewAI Yes

    A first-class terminal surface covers scaffolding, running, training, testing, replaying and deploying crews and flows, including login against enterprise SSO.

    Source retrieved 2026-09-13

  • Cursor Yes

    A terminal agent with interactive, plan, ask and print modes, sandbox controls and session resume.

    Source retrieved 2026-09-02

  • Deepgram Voice Agent API Qualified

    The dg CLI covers transcription, synthesis, text analysis and account management; its own getting-started enumeration and command reference contain no voice-agent command, while the agentic-tools page claims the CLI reaches voice agents. Both statements are recorded in notes.

    Source retrieved 2026-09-02

  • Devin Yes

    Devin CLI is a local terminal agent working on your own files, installed by curl script, Homebrew cask or Windows installer.

    Source retrieved 2026-09-02

  • Devin Desktop Qualified

    Devin Desktop installs a devin-desktop shell command for opening files, and its local agent shares the harness with Devin CLI, but the terminal agent is a separate product.

    Source retrieved 2026-09-03

  • Dify Yes

    Official difyctl CLI runs workspace apps from a terminal with OAuth sign-in and JSON output.

    Source retrieved 2026-09-02

  • Dust Yes

    A published npm CLI opens an interactive terminal chat with filesystem access, and takes single non-interactive messages that return JSON on stdout.

    Source retrieved 2026-09-09

  • ElevenLabs Agents Yes

    A standalone CLI authenticates by OAuth and pushes, pulls and diffs agent, tool and test configurations held as files in version control.

    Source retrieved 2026-09-02

  • Elicit Qualified

    There is no first-party Elicit CLI product; the vendor publishes a CLI tool among its API integration examples, so terminal use means running that sample against the Pro-and-above API.

    Source retrieved 2026-09-14

  • Faceless.so Qualified

    A published npm CLI covers the whole API with JSON output and exit codes built for agents, but it rides on API access, which the plan table grants only from Ultra upward.

    Source retrieved 2026-10-03

  • Factory Yes

    The Droid CLI installs via curl, Homebrew, npm or PowerShell and runs in any repository.

    Source retrieved 2026-09-02

  • Firecrawl Yes

    A published CLI logs in with an API key and runs scrape, search and interact commands from the terminal, alongside agent skills.

    Source retrieved 2026-09-24

  • GC AI Yes

    A published npm CLI covers the whole REST surface from a terminal, authenticating with an API key, so scripts and shell pipelines are a first-class path.

    Source retrieved 2026-09-14

  • Gemini Code Assist Yes

    Gemini CLI is included with both editions and shares its daily request quota with IDE agent mode; it is preinstalled in Cloud Shell.

    Source retrieved 2026-09-02

  • Genspark Qualified

    Claw exposes a browser terminal on its cloud machine and GenTeam is joined by pasting a connect command into a terminal, but Genspark ships no command-line client that drives its agents.

    Source retrieved 2026-09-15

  • GitHub Copilot Yes

    Copilot CLI is included on every plan, including Free, and can commit locally or on GitHub and delegate to coding agents.

    Source retrieved 2026-09-02

  • Glean Qualified

    No Glean CLI of its own; a plug-in exposes Glean tools inside terminal coding agents such as Claude Code, Codex, Gemini CLI and OpenCode.

    Source retrieved 2026-09-02

  • Google Agent Development Kit (ADK) Yes

    A terminal surface is first-class: an interactive run command with session save, resume and replay flags in the Python CLI, and an embedded launcher for Go binaries.

    Source retrieved 2026-09-15

  • goose Yes

    The CLI is a first-class surface with session, run, review, recipe, schedule, serve and terminal-integration commands, and shares provider and extension configuration with the desktop app.

    Source retrieved 2026-09-16

  • GPT Researcher Yes

    A documented cli.py takes the query, report type and tone as arguments and writes the finished report to an outputs directory.

    Source retrieved 2026-09-16

  • Gumloop Yes

    A first-party CLI runs agents, calls MCP tools, searches Company Brain and manages skills and artifacts from the terminal, and is included on both plans.

    Source retrieved 2026-09-30

  • HOL Guard Yes

    A Python CLI installs, initialises and health-checks Guard, and remains the documented alternative to the desktop app.

    Source retrieved 2026-09-29

  • Juggler Yes

    A headless command-line server is half of every download and runs on the machine holding the code; the same binary carries CLI subcommands for scaffolding, validating and linking extensions.

    Source retrieved 2026-09-17

  • Junie Yes

    The terminal is the primary surface, installed by a shell command on Linux, macOS and Windows, with interactive sessions, slash commands and shell passthrough.

    Source retrieved 2026-09-18

  • Kilo Code Yes

    Standalone CLI installed from npm, with sandboxed auto mode.

    Source retrieved 2026-09-02

  • LangChain Yes

    Several terminal surfaces exist: the LangGraph CLI for local dev servers, a Sandbox CLI on every plan, a LangSmith CLI for issues, and the dcode terminal coding agent.

    Source retrieved 2026-09-18

  • Langflow Yes

    A first-class CLI ships with the Python package and manages the server, API keys, migrations and superusers, though it is unavailable in the desktop build.

    Source retrieved 2026-09-25

  • LangGraph Yes

    langgraph-cli runs the Agent Server locally with langgraph dev, and builds, dockerises and deploys with build, dockerfile and deploy.

    Source retrieved 2026-09-02

  • Lead Scorer Qualified

    There is no Lead Scorer CLI of its own; terminal use means connecting the MCP server from Claude Code or Codex, for which the vendor publishes a plugin bundling the OAuth MCP and its skills.

    Source retrieved 2026-09-18

  • Letta Yes

    The primary surface is a terminal harness installed from npm, with headless mode for automation.

    Source retrieved 2026-09-02

  • LiveKit Agents Yes

    The lk CLI creates projects, deploys agents and manages LiveKit Cloud; agents can also be run and talked to locally in console mode.

    Source retrieved 2026-09-02

  • LlamaIndex Yes

    Skills and plugins install into terminal coding agents Claude Code and Codex; a Go CLI (llp) is a documented install path.

    Source retrieved 2026-09-02

  • Manus Qualified

    Through the desktop app the agent drives your machine's command line and the API exposes terminalExecute confirmations, but no Manus CLI client you type commands into is documented.

    Source retrieved 2026-09-19

  • Mastra Yes

    A Mastra CLI builds, starts and dev-serves the project (mastra build, mastra start, mastra dev).

    Source retrieved 2026-09-02

  • Microsoft Agent Framework Qualified

    A terminal surface exists only as developer tooling: the DevUI command line launches agents from a directory, and the Harness console apps are explicitly samples rather than shipped components.

    Source retrieved 2026-09-19

  • MindStudio Yes

    The npm client ships CLI commands to sync types, list agents and run a workflow from a terminal; the agent SDK additionally installs a standalone mindstudio binary that runs any action or an MCP server.

    Source retrieved 2026-09-02

  • Muse Code Yes

    The terminal is the primary surface: a native binary with a TUI, slash commands, live approvals and a command palette on macOS, Linux and Windows.

    Source retrieved 2026-09-20

  • n8n Yes

    Two terminals: n8n CLI over the API from any machine (workflow, execution, credential, project, package, audit commands) and a Server CLI on the host for backups and licences.

    Source retrieved 2026-09-02

  • OmniDimension Qualified

    The only terminal-run component is the local MCP server, installed and diagnosed with npx sub-commands; no CLI exists for building agents or dialling calls on its own.

    Source retrieved 2026-09-09

  • OpenAI Agents SDK (Python) Yes

    run_demo_loop gives an interactive terminal chat session with streaming and retained history; it is a demo/testing loop, not a packaged CLI binary.

    Source retrieved 2026-09-02

  • OpenAI Realtime API Qualified

    An official openai command-line tool exists for the API, but the documented CLI workflows stop at request/response calls; realtime sessions are driven from your own WebSocket, WebRTC or SIP code.

    Source retrieved 2026-09-02

  • OpenClaw Yes

    A global npm CLI plus a CLI-only installer give a full terminal surface, with subcommands for onboarding, gateway, agents, approvals, backup and MCP.

    Source retrieved 2026-09-21

  • OpenCode Yes

    The terminal is the primary surface: a TUI by default, plus scriptable subcommands for one-shot runs and server mode.

    Source retrieved 2026-09-21

  • OpenHands Yes

    A terminal CLI is included on every plan including the free local one, with headless and JSON output modes for scripting.

    Source retrieved 2026-09-21

  • Perplexity Yes

    A first-party pplx binary runs web search and page-snippet extraction from a terminal and returns JSON for pipelines.

    Source retrieved 2026-09-02

  • Pickaxe Yes

    A published CLI drives the whole account from a terminal, creating agents, syncing knowledge bases and configuring deployments, and is built for a coding agent to run on your behalf.

    Source retrieved 2026-09-09

  • PolyAI Yes

    The poly CLI pulls, branches, validates, pushes and chats against a project from the terminal.

    Source retrieved 2026-09-02

  • Pydantic AI Yes

    The clai command ships with the package and runs an interactive terminal chat against any model, and any agent object can launch the same interface with Agent.to_cli_sync().

    Source retrieved 2026-09-21

  • Relevance AI Qualified

    No first-party CLI of its own, but a published Claude Code plugin bundles the MCP server and skills so agents can be built and managed from a terminal.

    Source retrieved 2026-09-22

  • Replit Agent Qualified

    A full Git command line is available through the workspace Shell, and terminal MCP clients can drive Agent, but there is no local Replit CLI.

    Source retrieved 2026-09-22

  • Salesforce Agentforce Qualified

    Salesforce CLI drives the whole agent lifecycle from a terminal, but the Vibes coding agent itself is distributed only as a VS Code extension and a hosted IDE. There is no standalone Vibes CLI.

    Source retrieved 2026-09-03

  • SalesTouch Yes

    The documented default connection path is the Codex command-line client, with the graphical Codex app given as the alternative.

    Source retrieved 2026-09-25

  • SEOmatic Yes

    An install-less npm CLI audits any URL and analyses crawl logs with no account, and drives the REST API with a key once you have one.

    Source retrieved 2026-10-03

  • Sim Yes

    A `sim` CLI runs workflows, queries tables and reads logs, with a JSON output mode for scripting.

    Source retrieved 2026-09-03

  • SkaleAgents Qualified

    There is no SkaleAgents CLI of its own, but terminal clients are documented: Claude Code and Codex are configured and re-authenticated with their own shell commands.

    Source retrieved 2026-10-03

  • StackAI Qualified

    The published CLI installs and maintains the self-hosted platform on customer servers; it is an operations tool for on-prem deployments, not a way to drive an agent from a terminal.

    Source retrieved 2026-09-09

  • Tale Qualified

    A Tale CLI exists for deployment and configuration projects, and terminal agents can call the REST and MCP endpoints, but agent runs themselves are driven from the web app.

    Source retrieved 2026-10-03

  • Telnyx Voice AI Agents Yes

    Two terminal surfaces exist: a full resource-level Telnyx CLI and an Agent CLI whose composite commands stand up an assistant on a phone number in one invocation.

    Source retrieved 2026-09-09

  • TinyFish Yes

    A CLI ships alongside the SDKs and MCP endpoint and can start a run from a goal.

    Source retrieved 2026-09-03

  • UiPath Agent Builder Yes

    A dedicated CLI drives the low-code agent projects themselves, not just the Python path: it scaffolds, validates, packs, pushes back into the Agent Builder UI, deploys and runs them, and is separate from the coded-agent tool.

    Source retrieved 2026-09-03

  • Undermind Qualified

    There is no Undermind CLI of its own; terminal use comes through agent CLIs such as Codex and Claude Code, which you configure with the MCP endpoint.

    Source retrieved 2026-09-26

  • Unify Qualified

    There is no Unify CLI of its own; terminal use happens only by installing the Unify plugin inside Claude Code.

    Source retrieved 2026-09-27

  • Vapi Yes

    The Vapi CLI covers assistants, numbers, calls, campaigns and logs from the terminal.

    Source retrieved 2026-09-03

  • Vecbase Yes

    There is a real terminal surface: a vecbase CLI installed by a shell script, with subcommands for update, per-client setup and browser-based auth login.

    Source retrieved 2026-09-27

  • Vellum Yes

    A first-class `vellum` CLI sends messages, tails the event stream and manages processes, and is also how self-hosted runtimes are started and inspected.

    Source retrieved 2026-09-27

  • VeraDial Qualified

    There is no VeraDial CLI, but developer terminal clients can attach to the read-only MCP server after a manual one-time approval, which gives terminal access to account records only.

    Source retrieved 2026-09-22

  • Warmly Qualified

    Claude Code and other MCP-compatible terminal agents can call the three tools; there is no Warmly CLI.

    Source retrieved 2026-09-03

  • Warp Yes

    Two terminal surfaces: the Warp Terminal app for macOS, Linux and Windows, and a standalone agent CLI that runs in any terminal without the app.

    Source retrieved 2026-09-03

  • Webhound Qualified

    There is no Webhound CLI; the terminal surface is borrowed from CLI agents such as Claude Code and Codex that call it over MCP.

    Source retrieved 2026-09-27

  • WTAgent Yes

    The terminal is the only interface: you install globally with npm, run wtagent in a project directory and type tasks at a prompt.

    Source retrieved 2026-09-21

  • Xquik Yes

    A CLI is published alongside the language SDKs, and terminal agents such as Claude Code and Gemini CLI connect over Streamable HTTP MCP.

    Source retrieved 2026-10-03

  • Zed Qualified

    The zed CLI opens files and can be driven from scripts, but it launches the editor rather than running an agent headlessly; agent CLIs instead run as Terminal Threads inside Zed.

    Source retrieved 2026-09-29

  • Zencoder Qualified

    Terminal agents are supported as runtimes — its own Zen CLI plus Claude Code and Codex — but the docs frame them as hosted inside the editor rather than as a standalone Zencoder terminal product.

    Source retrieved 2026-09-09

  • zot Yes

    The terminal is the product: a single static binary dropped on your PATH gives an interactive TUI plus one-shot print and streaming invocations.

    Source retrieved 2026-09-28

The population

143 listings in this index are Published. This is the population as of 2026-10-03. 143 of the 143 published listings in this index carry a settled answer for “On the command line”. That row asks: Is there a terminal or CLI surface? Every listing in the index is settled on this row, so nothing is left out.

Findings

  1. 67 of the 143 settled listings answer “Yes” for “On the command line”. That is 47% of the settled set.

  2. 30 of the 143 settled listings answer “Qualified” for “On the command line”. That is 21% of the settled set.

  3. 17 of the 143 settled listings answer “No” for “On the command line”. That is 12% of the settled set.

  4. 25 of the 143 settled listings answer “Does not apply” for “On the command line”. That is 17% of the settled set.

  5. 4 of the 143 settled listings answer “Not established” for “On the command line”. That is 3% of the settled set.

1 of those yes answers is LlamaIndex, which still answers yes for “On the command line”. Its listing carries the stored answer and links to the source recorded for it. This is one worked example, not an independent audit of every source in the census.

What we counted, and how

Each figure above is a count over the “On the command line” row of the listing datasheet, taken from the same stored answer the listing page renders. The denominator is the 143 listings whose answer is settled, meaning one of yes, qualified, no, does not apply, not established. On this row that is the whole index, because every listing has a recognised stored answer. Every number here is stored with the read-only query that reproduces it and re-run every sixty seconds against the live corpus, so a figure that stops reproducing surfaces as drift rather than as a stale sentence nobody notices. Our full method covers how a datasheet row is settled in the first place.

Limitations

This counts stored datasheet answers, not independently tested capabilities. A sourced answer can record a vendor statement or our reading of published evidence; this census does not re-fetch those sources. “Not established” means we have not established an answer. That can reflect vendor nondisclosure, blocked evidence, or unfinished research, not a no. An absent or unrecognised answer is excluded rather than treated as a researched finding. The figures are restated when the stored corpus changes; the date above is the count used for this published version, not a new verification of the vendors.

Get the next report

New agents rankings and fresh data reports. One short email, one-click unsubscribe.