Skip to content

The datasheet for every AI agent

Research report · data-report

Ships it: a census of 143 listings

18 of 143 listings answer yes for “Ships it”. Counted from our own datasheet, with the query that reproduces every figure.

Oct 1, 2026updated Oct 3, 20262 min readSource-linked research

Ships it, across 143 settled listings

  1. Yes18 of 143

  2. Qualified52 of 143

  3. No10 of 143

  4. Does not apply61 of 143

  5. Not established2 of 143

Of the 143 Published listings whose datasheet settles “Ships it”, the largest group is “Does not apply” at 61 (43%). Measured 2026-10-03.

Denominator: 143 Published listings whose datasheet settles “Ships it”.Method: Counted from the stored datasheet answer for “Ships it” 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/ships-code-census/" target="_blank" rel="nofollow noopener"><img src="https://theagentsindex.com/charts/ships-code-census.png" alt="Ships it, across 143 settled listings — The Agents Index" title="Ships it, across 143 settled listings — The Agents Index" width="640" loading="lazy" style="max-width:100%;height:auto" /></a>

Explore the matching agents

Ships it · Datasheet criterion ships_code · 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 70 current matching records
  • Ada Qualified

    Only for Ada's own sandboxed Code tools: promoting a change set puts the Python live in production conversations and a promote can be reverted, but a human approves and nothing deploys to your infrastructure.

    Source retrieved 2026-09-09

  • Agentix Labs Yes

    The firm takes engagements through to live operation rather than stopping at a prototype, deploying production agents with logging, approvals and rollback paths.

    Source retrieved 2026-09-29

  • AgentsKit.js Qualified

    There is no deploy feature; a GitHub Actions integration can trigger workflow_dispatch and the allowlisted shell can run git, so shipping is a pipeline you wire yourself.

    Source retrieved 2026-09-09

  • Agno Qualified

    Publishing is deliberately outside the default loop; commits, PRs and deploys must be exposed as separately gated tools.

  • Aident Loadout Qualified

    It gives coding agents actions against deployment platforms, so a deploy can be triggered through Loadout, but Loadout itself produces no diff and the vendor documents no release pipeline of its own.

    Source retrieved 2026-10-03

  • Aisera Agent Composer (Agent Studio) Yes

    Built agents and workflows are deployed by the builder itself: publishing is a mandatory, version-controlled step in the Admin UI with rollback, so work does not stop at a draft.

    Source retrieved 2026-09-24

  • Antigravity CLI Qualified

    The agent executes local shell commands under your permission rules, so it can run deploy scripts you authorise, but no release, CD or environment-promotion feature is documented anywhere in the CLI docs.

    Source retrieved 2026-09-10

  • Arsum Yes

    Deployment is inside the engagement rather than left to the client, and the handover step names the deployment environment and rollback explicitly.

    Source retrieved 2026-10-03

  • Augment Code Qualified

    Agents drive a change through CI to merge; the vendor does not claim the agent deploys or releases.

    Source retrieved 2026-09-02

  • AutoGen Qualified

    A team can execute the code it writes in a shell or Docker container, but nothing in the docs claims it opens pull requests or deploys to an environment.

    Source retrieved 2026-09-19

  • Bitsclan IT Solutions Yes

    Delivery runs past the diff into production: they publish a CI/CD-based deployment process, and the agent engagement has its own launch and post-launch tuning stage.

    Source retrieved 2026-09-11

  • Bland Qualified

    It ships agent versions to live phone traffic with canary, promotion and rollback; it does not deploy application code.

  • Bolt Yes

    Bolt deploys the finished project itself to a live bolt.host address, with custom domains and Netlify as alternatives on paid plans.

    Source retrieved 2026-09-11

  • Botpress Qualified

    One CLI command ships an ADK agent to a production bot on Botpress Cloud; it does not open a pull request or deploy to infrastructure you own.

    Source retrieved 2026-09-02

  • Browserbase Yes

    Browser automations are deployed as hosted Functions, published and invoked from the CLI or API.

  • Cartesia Qualified

    The CLI deploys the agent code you wrote to Cartesia's cloud; the product itself produces no diffs to ship.

  • Claude Agent SDK Qualified

    Bash gives it git and script execution, so a deploy command is reachable, but the SDK documents no deployment surface of its own.

    Source retrieved 2026-09-02

  • Claude Code Qualified

    The claimed workflow ends at a commit, branch and pull request; deployment is only whatever shell command Claude is allowed to run.

    Source retrieved 2026-09-02

  • Cline Qualified

    It goes past the diff by running your own commands and by committing or opening a PR from the Kanban board, but Cline ships no deployment service of its own.

    Source retrieved 2026-09-12

  • CodeRabbit Qualified

    Accepted work lands in Git as a commit on the PR branch or a stacked pull request, pushed with your own credentials; nothing in the docs deploys or releases the change.

    Source retrieved 2026-09-15

  • Codex CLI Qualified

    Locally it stops at edits plus commands you allow; pull requests are opened from Codex cloud or the GitHub Action.

    Source retrieved 2026-09-02

  • Crescendo Qualified

    It ships what it builds only into its own runtime: a tested integration is published as a governed tool the assistant can call in live chat and voice, with no deployment to customer infrastructure.

    Source retrieved 2026-09-12

  • Cresta Qualified

    Cresta deploys the agents it builds through governed rollouts, A/B comparison against a baseline and one-click rollback; it does not ship application code to your own environments.

    Source retrieved 2026-09-02

  • CrewAI Qualified

    It ships the automation itself rather than landing code changes in your repository: crewai deploy create pushes a crew or flow to AMP.

  • Cursor Qualified

    It stops at the branch and pull request; deployment stays with your own CI, though a CLI run in CI can be given git and gh access.

    Source retrieved 2026-09-02

  • Devin Qualified

    Stops at the pull request in your repository; your team merges and ships through its existing pipeline.

    Source retrieved 2026-09-02

  • Devin Desktop Qualified

    One-click deploys exist but only through the legacy Cascade agent, only to Netlify, in beta, and capped at 1 deploy/day on Free and 10 on Pro; agents otherwise stop at a pull request.

    Source retrieved 2026-09-03

  • ElevenLabs Agents Qualified

    The CLI pushes agent configuration to the platform and into CI/CD; what it deploys is agents, not application code.

  • Factory Qualified

    High autonomy permits pushes, deployments and migrations; the Release stage that tracks ship gates is Private Preview.

    Source retrieved 2026-09-02

  • Gemini Code Assist Qualified

    One documented deploy path: the /deploy slash command in VS Code agent mode, which needs Cloud Run enabled and the Cloud Run MCP server installed as a Gemini extension. Not supported in JetBrains IDEs.

    Source retrieved 2026-09-02

  • Genspark Yes

    Deployment is claimed as the last step of the agent's own loop, so it goes past producing a diff; Claw can additionally host a site on its cloud computer.

    Source retrieved 2026-09-15

  • Google Agent Development Kit (ADK) Yes

    Deployment is part of the framework: one command pushes an agent to Google Cloud, and the same agent can be containerised and run on infrastructure you control instead.

    Source retrieved 2026-09-15

  • goose Qualified

    goose has no managed deployment pipeline of its own; it ships work by executing your shell and git commands, and the vendor positions CI/CD build and deployment workflows as tasks you script around the CLI.

    Source retrieved 2026-09-16

  • Gumloop Qualified

    Gumloop builds and deploys custom MCP servers straight from a GitHub repository on every push to main, but that ships the connector code, not the buyer's own applications.

    Source retrieved 2026-09-09

  • Juggler Qualified

    The agent edits files and executes shell commands, but no deployment or release capability is described.

  • Julius AI Qualified

    It goes past the artefact: generated websites and web apps publish to a public Julius link on every plan, but the vendor states custom domains and production backends need outside hosting.

    Source retrieved 2026-09-17

  • Junie Qualified

    It reaches further than a diff, committing, pushing branches and opening pull requests through the GitHub Action, but no deployment or release step is documented anywhere on the site.

    Source retrieved 2026-09-18

  • Kilo Code Qualified

    Cloud Agents commit and push to a session branch and raise pull requests; the vendor names deployments only in a comparison claim, and documents no deploy pipeline.

    Source retrieved 2026-09-02

  • LangChain Yes

    LangSmith Deployment runs agents in production with one-click deploy from Studio and a control-plane API; the free Developer plan shows Deployment as N/A, so shipping starts at Plus.

    Source retrieved 2026-09-18

  • Letta Qualified

    Commits, pushes and opens pull requests; it cannot approve PRs and no deployment step is claimed.

    Source retrieved 2026-09-02

  • Lindy Qualified

    Lindy ships what it builds to its own hosted lindy.link URL and rebuilds on the same address, so work reaches an audience without a deploy step, but it does not deploy into your own infrastructure.

    Source retrieved 2026-09-19

  • LlamaIndex Qualified

    What ships is the agent workflow itself: single-click deploy straight from a GitHub repo, listed on all four plans.

  • Manus Yes

    Deploys what it builds on managed infrastructure with a custom domain, so it goes past the diff; the API exposes the same publish step and deploys require a confirmation.

    Source retrieved 2026-09-19

  • Microsoft Agent Framework Qualified

    The beta agent-framework-tools package gives an agent local or Docker shell execution, so it can run any command you approve; no deploy, release or pull-request step is documented.

  • MindStudio Yes

    Publishing and deploying agents is unlimited on both paid tiers; deployment targets include hosted web apps, website embeds, webhooks, scheduled runs and, on Business, custom domains.

    Source retrieved 2026-09-02

  • Moveworks Qualified

    It deploys the AI agents built in Agent Studio, including promotion between instances; it ships no application code.

  • Muse Code Qualified

    It can run shell commands including git operations when you ask, but by design it stops short of committing or pushing on its own, and no deployment surface is documented.

    Source retrieved 2026-09-20

  • n8n Qualified

    What it ships is the automation itself: publishing puts a workflow or agent version into production and schedules or channels then run it. It does not deploy application code.

    Source retrieved 2026-09-02

  • Octacer Yes

    Delivery runs past the diff into production: they take the platform live and then monitor and harden it inside the client's operation.

    Source retrieved 2026-09-26

  • OpenClaw Qualified

    A Gateway publication broker creates commits, pushed branches and pull requests from a session; nothing in the docs claims it deploys.

    Source retrieved 2026-09-02

  • OpenCode Qualified

    It stops at a branch and pull request in your GitHub Actions runner; no deployment step is documented anywhere on the site.

    Source retrieved 2026-09-21

  • OpenHands Qualified

    It ends at a pull request rather than a deploy: the documented automation loop opens a PR and leaves the merge and release to you.

    Source retrieved 2026-09-21

  • Perplexity Qualified

    Through the API the agent pushes a branch and opens or updates a pull request; no deployment step or target is documented.

    Source retrieved 2026-09-02

  • PolyAI Qualified

    Agent versions ship through Draft, Sandbox, Pre-release and Live with rollback, drivable from the CLI or Agents API, but promotion to Live is always a manual step.

    Source retrieved 2026-09-02

  • Pydantic AI Qualified

    There is no deployment feature; the Coder shell tool runs unrestricted commands that outlive the run, so an agent can execute deploy commands you allow, and the docs warn to sandbox untrusted work.

    Source retrieved 2026-09-21

  • Replit Agent Yes

    Agent does not stop at a diff: publishing puts a running instance on Replit's cloud, with custom domains and monitoring included.

    Source retrieved 2026-09-22

  • SafeNet Creations Yes

    Delivery goes past the build to deployment of the accepted scope, with monitoring and access documented at launch.

    Source retrieved 2026-09-04

  • Salesforce Agentforce Yes

    It deploys metadata to the org itself, and through the Salesforce DX MCP Server's DevOps Center tools it commits and opens pull requests; deploy and delete stay approval-gated by default.

    Source retrieved 2026-09-03

  • SEOmatic Qualified

    It does publish live: agents write straight into the connected CMS, and for sites built in code the fix arrives as a GitHub pull request that you merge yourself, so on that surface it stops at the diff.

    Source retrieved 2026-10-03

  • Sierra Qualified

    It ships agent versions rather than code: publishing to ChatGPT can run through CI/CD, but nothing deploys application software.

    Source retrieved 2026-09-26

  • Sim Qualified

    Deploys what it builds: workflows ship as versioned REST endpoints, with deploy/undeploy/rollback callable from CI. What ships is workflows, not code.

    Source retrieved 2026-09-03

  • TecAdRise Yes

    Delivery does not stop at handing over artefacts: they put builds into production on client or VPS infrastructure and keep monitoring and iterating after go-live.

    Source retrieved 2026-09-12

  • UiPath Agent Builder Yes

    It does not stop at a design: the agent is published and deployed as an Orchestrator process, the CLI provisions a serverless machine and robot user for it, and Orchestrator then schedules and monitors the runs.

    Source retrieved 2026-09-03

  • Vecbase Qualified

    Delivers a live preview plus deployable source files; nothing on vecbase.com claims it deploys to your own hosting or merges to your repo.

    Source retrieved 2026-09-03

  • Warp Qualified

    Agents commit, push branches and open pull requests through the GitHub integration, but no page describes the agent performing a deployment.

    Source retrieved 2026-09-03

  • WTAgent Qualified

    Pushing, releasing and deploying are permitted but classified as external-risk actions that always stop for an explicit human approval prompt.

    Source retrieved 2026-09-21

  • Zed Qualified

    There is no deployment feature; the agent stops at a reviewable diff, though its terminal tool can execute any shell command you permit, which is how a deploy script would run.

    Source retrieved 2026-09-29

  • Zencoder Qualified

    It commits, pushes the task branch and opens a pull request with a pre-filled description, and can auto-push, but no page claims it deploys to an environment.

    Source retrieved 2026-09-09

  • zot Qualified

    There is no deployment feature; the agent stops at edits on disk and whatever your own shell commands do, since git and deploy scripts run through the generic bash tool.

    Source retrieved 2026-09-28

  • ZTABS Yes

    Delivery runs past the diff into production: they build CI/CD pipelines that deploy to staging and production, and deploy agents into the client environment with monitoring.

    Source retrieved 2026-09-25

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 “Ships it”. That row asks: Can it deploy, or does it stop at the diff? Every listing in the index is settled on this row, so nothing is left out.

Findings

  1. 18 of the 143 settled listings answer “Yes” for “Ships it”. That is 13% of the settled set.

  2. 52 of the 143 settled listings answer “Qualified” for “Ships it”. That is 36% of the settled set.

  3. 10 of the 143 settled listings answer “No” for “Ships it”. That is 7% of the settled set.

  4. 61 of the 143 settled listings answer “Does not apply” for “Ships it”. That is 43% of the settled set.

  5. 2 of the 143 settled listings answer “Not established” for “Ships it”. That is 1% of the settled set.

1 of those yes answers is ZTABS, which still answers yes for “Ships it”. 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 “Ships it” 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.