# CLIKA documentation

> ClikaRT is the inference runtime, ClikaRT CLI the command line interface that infers, serves and benchmarks popular models on it (built on Modelverse, the CLIKA model library), and the Platform the service that runs benchmarks on devices and issues the runtime licenses. Every page below is a plain address on this site; the site also offers a self-contained offline copy (the "Offline Docs" button).

Every address below is the page as Markdown; the same address without `.md` is the page as HTML. `https://docs.clika.io/llms-full.txt` is every page listed here as one text, in this order.

## ClikaRT

- [ClikaRT](https://docs.clika.io/clikart.md): ClikaRT is CLIKA's inference runtime, a C++ library with Python and Kotlin bindings, that runs models on CPUs, GPUs and other hardware accelerators through one public API.
- [API reference](https://docs.clika.io/clikart/api.md): The ClikaRT public C++ API, by namespace and class.
- [Kotlin API](https://docs.clika.io/clikart/api-kt.md): The ClikaRT Kotlin binding, documented from its sources.
- [clika_runtime](https://docs.clika.io/clikart/api-py.md): The clika_runtime module, introspected from the installed wheel.
- [Language bindings](https://docs.clika.io/clikart/bindings.md): ClikaRT has two API levels: FULL, the whole public surface, for C++ and Python; INFERENCE, everything that runs a ready model, for Kotlin. Every binding compiles against the C++ public headers.
- [Additional examples](https://docs.clika.io/clikart/examples.md): The example programs that ship with a release: the runtime's own CMake projects inside every desktop archive, and the examples archive with complete C++, Python, Kotlin and Android programs beside their recorded output.
- [First steps](https://docs.clika.io/clikart/getting-started.md): New to ClikaRT? Start here. What the runtime is, how to install it, and the tutorial series.
- [Complete installation](https://docs.clika.io/clikart/getting-started/complete-installation.md): Per-platform toolchains, IDE setup, and the full failure catalog. Under construction.
- [Deploy to mobile](https://docs.clika.io/clikart/getting-started/first-program/deploy-to-mobile.md): Cross-compile the part 4 program for Android, push it over adb, and run it on the phone's GPU. The code does not change.
- [Devices and the async model](https://docs.clika.io/clikart/getting-started/first-program/devices-and-async.md): Discover the machine's devices, place tensors on them, and understand dispatch vs ready, and why host reads are always safe.
- [Serve it](https://docs.clika.io/clikart/getting-started/first-program/serve-it.md): Wrap part 4's model in the serving runtime: a schema, a lambda serving the forward pass, and an executor answering requests.
- [Tensors and operators](https://docs.clika.io/clikart/getting-started/first-program/tensors-and-operators.md): Factories and dtypes, the ops:: library, views, operator sugar, and reading values back to the host.
- [Your first pipeline](https://docs.clika.io/clikart/getting-started/first-program/your-first-pipeline.md): Load weights and input from disk with ClikaRT::io, compute on the best device, read the results back.
- [Your first program](https://docs.clika.io/clikart/getting-started/first-program/your-first-program.md): Link the bundle, print the version, make your first tensor, and meet the error model.
- [Get ClikaRT](https://docs.clika.io/clikart/getting-started/get-clikart.md): Pick your OS, architecture and accelerator, and get the download, verify and extract commands for the archive the CLIKA platform hands out.
- [Quick install](https://docs.clika.io/clikart/getting-started/installation.md): Install a C++ toolchain, extract the ClikaRT bundle, and prove it works in two commands.
- [What to read next](https://docs.clika.io/clikart/getting-started/next-steps.md): Where the documentation goes after the tutorial: how-to guides, the API reference, Modelverse, system requirements.
- [ClikaRT at a glance](https://docs.clika.io/clikart/getting-started/overview.md): What the ClikaRT runtime is and is not, and the five-minute mental model.
- [How-to guides](https://docs.clika.io/clikart/how-to.md): Problem-oriented recipes. Each guide answers one "how do I X" with a worked example, verified against the bundle.
- [Author a model in Python](https://docs.clika.io/clikart/how-to/author-a-model-in-python.md): Write a Llama-class decoder as clika_runtime.nn modules, load a Hugging Face checkpoint as one resident copy, fuse its projections, decode through a KV cache with a one-token lookahead, and measure it against the C++ command line.
- [Coming from PyTorch or Hugging Face](https://docs.clika.io/clikart/how-to/coming-from-pytorch.md): The conventions a reader who knows PyTorch and the Hugging Face libraries meets first: channels-last tensors and OHWI weights, integer slots, reads that wait, the JSON and audio accessors, the tokenizer's inputs, and the generate calls side by side.
- [Control asynchronous execution](https://docs.clika.io/clikart/how-to/control-async-execution.md): Know when dispatched work actually runs, get completion callbacks, and force synchronous or lazy execution with the stream scopes.
- [Edit a graph](https://docs.clika.io/clikart/how-to/edit-a-graph.md): Change a ModelGraph before finalize(): rewire the reads of a value, change what a node reads, rename a node, and edit the graph's inputs and outputs. A refused edit changes nothing; then optimize, finalize and run the edited graph.
- [Add ClikaRT to an existing CMake project](https://docs.clika.io/clikart/how-to/existing-cmake-project.md): Link the bundle into a project that already builds, with find_package or add_subdirectory, and the one linker flag serving-runtime consumers need.
- [Export a graph to ONNX](https://docs.clika.io/clikart/how-to/export-a-graph-to-onnx.md): Write a ModelGraph as an ONNX model file, or build the same model in memory: read it back through OnnxModel::open and compile, keep a named dynamic dimension, choose where the weights go and the opset, and meet what the export refuses.
- [Handle errors by code](https://docs.clika.io/clikart/how-to/handle-errors-by-code.md): Read a failure's three channels, branch on the stable code name, and switch policy on the coarse status, in every language.
- [License the runtime](https://docs.clika.io/clikart/how-to/license-the-runtime.md): Where ClikaRT and Modelverse read the credential your platform issues, per language and packaging, and what a refused call looks like.
- [Load images and audio for inference](https://docs.clika.io/clikart/how-to/load-images-and-audio.md): Decode image files into [H, W, C] tensors and audio into resampled waveforms, then make them model-ready with ops.
- [Load quantized weights from a GGUF file](https://docs.clika.io/clikart/how-to/load-quantized-weights.md): Read a block-quantized checkpoint, keep the weights packed, and serve them through QLinearWoQ at their on-disk footprint.
- [Merge and split](https://docs.clika.io/clikart/how-to/merge-and-split.md): Join ModelGraphs into one graph that runs them all, side by side, with the outputs of one feeding the inputs of another, or across two devices; cut a graph into parts at its cheapest cut or by a function, extract the operators between values, and merge the parts back.
- [Optimize and finalize](https://docs.clika.io/clikart/how-to/optimize-and-finalize.md): A traced or compiled ModelGraph comes back as built. Run the graph optimizer with optimize() and read its report, choose its transforms and its defaults, place the graph with to(), and finalize() it to run.
- [Package ClikaRT in an Android app](https://docs.clika.io/clikart/how-to/package-clikart-in-an-android-app.md): What an Android app ships to run the runtime and the model library on the phone: the one artifact and the libraries inside it, the build requirements, the license credential, the thread count, the cache root, and a server that outlives the screen.
- [Preprocess inputs with processors](https://docs.clika.io/clikart/how-to/preprocess-with-processors.md): Model-specific image and audio preprocessing: build a resize/rescale/normalize pipeline or a log-mel front end from knobs, or load it from the model's own config.
- [Structure inputs and outputs as pytrees](https://docs.clika.io/clikart/how-to/pytrees.md): Flatten and rebuild nested containers of tensors, name every leaf by its key path, register your own classes and dataclasses as nodes, serialize a structure, and pass whole trees through compile, trace, eval and save.
- [Query a graph](https://docs.clika.io/clikart/how-to/query-a-graph.md): Read a ModelGraph as data: its nodes, values and edges, search by name or predicate, walks and topological orders, paths and the critical path, dominance, regions, cones and the cheapest cut, views across edits, and networkx.
- [Run an ONNX model](https://docs.clika.io/clikart/how-to/run-an-onnx-model.md): Open an ONNX file (or build one from scratch), compile it into a ModelGraph, optimize and finalize it, and execute it by position or by name.
- [Serve a model over HTTP](https://docs.clika.io/clikart/how-to/serve-over-http.md): Put compute behind HTTP endpoints with the built-in server, answer JSON requests with tensor results, and stream tokens with server-sent events.
- [Tokenize text and apply a chat template](https://docs.clika.io/clikart/how-to/tokenize-and-chat-templates.md): Turn text into token ids and back, align tokens to source bytes, render a conversation with the model's own chat template, and batch for a model.
- [Trace eager code to graphs](https://docs.clika.io/clikart/how-to/trace-eager-code-to-graphs.md): Capture an eager function or module as a ModelGraph with trace, inspect it, optimize, finalize and run it, and wrap hot paths in compile for capture-and-replay.
- [Use ClikaRT from Python](https://docs.clika.io/clikart/how-to/use-clikart-from-python.md): The clika-runtime wheel: NumPy in and out with explicit copy semantics, math that reads as math, modes as strings, models as nn.Module, and errors typed by class.
- [Use ClikaRT with PyTorch](https://docs.clika.io/clikart/how-to/use-clikart-with-pytorch.md): Run a torch.compile model on the runtime with backend=\"clika\", move tensors across the two libraries over DLPack without a copy, and convert an eager torch.nn.Module tree into runtime layers.
- [Wrap existing memory without copying](https://docs.clika.io/clikart/how-to/wrap-existing-memory.md): Tensor::from_blob over buffers your application already owns: the borrow and adopt contracts, strided views, and what happens on a device move.
- [Write a custom operator](https://docs.clika.io/clikart/how-to/write-a-custom-operator.md): Subclass nn::Module two ways: compose built-ins into a reusable block, or run your own kernel through the runtime with output_shapes and compute.
- [Write a transform](https://docs.clika.io/clikart/how-to/write-a-transform.md): Write graph transforms of your own: functions over the ModelGraph edits that optimize() runs beside the runtime's transforms, operators built with add_node, insert and replace, and rewrite rules that replace each occurrence of a Pattern.
- [System requirements](https://docs.clika.io/clikart/system-requirements.md): Platforms, accelerators, host software, memory and storage for running ClikaRT.

## ClikaRT CLI

- [ClikaRT CLI](https://docs.clika.io/modelverse.md): ClikaRT CLI is the command line interface that infers, serves and benchmarks popular models, built on top of Modelverse, the CLIKA model library, and running them on ClikaRT.
- [Additional examples](https://docs.clika.io/modelverse/examples.md): The Modelverse example programs in the release's examples archive, standalone CMake projects against the installed package, from the catalog probe to vision, OCR, reranking and vision-language chat.
- [First steps](https://docs.clika.io/modelverse/getting-started.md): New to Modelverse? Start here. What the model library is, how to install it, and your first model from catalog to served endpoint.
- [Serve and chat](https://docs.clika.io/modelverse/getting-started/first-model/chat-and-serve.md): Host the model behind an OpenAI-compatible HTTP endpoint, hold a conversation on the built-in chat page, and call it with curl.
- [Fetch and prompt](https://docs.clika.io/modelverse/getting-started/first-model/fetch-and-prompt.md): Download the model snapshot, generate your first text, and control sampling from the command line.
- [Pick a model](https://docs.clika.io/modelverse/getting-started/first-model/pick-a-model.md): Read the catalog, resolve a model's identity, and see what a download would cost, all before any weights move.
- [Use it from your code](https://docs.clika.io/modelverse/getting-started/first-model/use-it-from-code.md): Link the Modelverse library and run the same model inside your own program: snapshot, load, pipeline, text.
- [Get Modelverse](https://docs.clika.io/modelverse/getting-started/get-modelverse.md): Pick your platform's ClikaRT archive; Modelverse ships inside it, with the download and verify commands.
- [Quick install](https://docs.clika.io/modelverse/getting-started/installation.md): Extract the release archive and prove Modelverse works in two commands, no toolchain required.
- [What to read next](https://docs.clika.io/modelverse/getting-started/next-steps.md): Where the documentation goes after the tutorial: how-to guides, model requirements, the examples, and the ClikaRT docs underneath.
- [Modelverse at a glance](https://docs.clika.io/modelverse/getting-started/overview.md): What the Modelverse model library is and is not, and the five-minute mental model.
- [A model from scratch](https://docs.clika.io/modelverse/getting-started/own-model/a-model-from-scratch.md): Write an embedding model's checkpoint by hand (config, tokenizer, random weights), load it through the registry, and compare two texts, all offline.
- [Your architecture joins the catalog](https://docs.clika.io/modelverse/getting-started/own-model/join-the-catalog.md): Rename the toy model's architecture so nothing matches it, then register a family of your own and watch the same directory resolve.
- [Make it embed](https://docs.clika.io/modelverse/getting-started/own-model/make-it-embed.md): Give the my-model family the smallest real embedding backend, opt into the similar and serve commands, and compare texts through your own family.
- [How-to guides](https://docs.clika.io/modelverse/how-to.md): Problem-oriented recipes. Each guide answers one "how do I X" with a worked session and its real output.
- [Add your own node to a model pipeline](https://docs.clika.io/modelverse/how-to/add-a-pipeline-node.md): Compose custom pre- or post-processing with the zoo's serving nodes in one pipeline, on the same request surface.
- [Benchmark a model on this machine](https://docs.clika.io/modelverse/how-to/bench-a-model.md): The bench flow: serving-shaped sweep cells, the structured report, and how to read its columns before committing a deployment.
- [A conversational AI as one pipeline](https://docs.clika.io/modelverse/how-to/conversational-ai-pipeline.md): Compose speech-to-text, a chat model and text-to-speech into a single runtime::Pipeline: wav bytes in, the spoken answer out.
- [Estimate depth, segment and detect](https://docs.clika.io/modelverse/how-to/depth-segment-detect.md): The three image verbs: per-pixel depth maps, semantic segmentation masks, and open-vocabulary detection, from the command line and over HTTP.
- [Embed, compare and rerank](https://docs.clika.io/modelverse/how-to/embed-and-rerank.md): Dense vectors, similarity matrices and cross-encoder reranking from the command line and over HTTP: the retrieval building blocks.
- [Add Modelverse to an existing CMake project](https://docs.clika.io/modelverse/how-to/existing-cmake-project.md): find_package against the installed directory, staging the runtime beside your binary, and picking the dist on cross builds.
- [Register your own model family](https://docs.clika.io/modelverse/how-to/register-your-own-family.md): Teach the catalog an architecture it does not know, from the snapshot layout through identity matching to runnable commands.
- [Run a specific GGUF quantization](https://docs.clika.io/modelverse/how-to/run-a-gguf-quantization.md): Pick one weight option of a repo that ships many, by tag, glob or exact file, and see the size table before anything downloads.
- [Run fully offline](https://docs.clika.io/modelverse/how-to/run-fully-offline.md): Fetch on a connected machine, move the model directory, and make any network touch an error instead of a surprise.
- [Script clikart-cli](https://docs.clika.io/modelverse/how-to/script-the-cli.md): The contracts that make clikart-cli scriptable: exit codes, the stdout/stderr split, option templates and shell completion.
- [Serve a model over the OpenAI and Anthropic APIs](https://docs.clika.io/modelverse/how-to/serve-openai-compatible.md): Host any runnable model behind the built-in HTTP server, stream completions, and point existing OpenAI and Anthropic clients at it.
- [Speak text](https://docs.clika.io/modelverse/how-to/speak-text.md): Text-to-speech from the command line and over HTTP, with voice references and the synthesis knobs.
- [Transcribe audio](https://docs.clika.io/modelverse/how-to/transcribe-audio.md): Speech-to-text with a Whisper model, from the command line and as an OpenAI-compatible HTTP endpoint.
- [Translate text](https://docs.clika.io/modelverse/how-to/translate-text.md): Neural machine translation from the command line and over HTTP, with the model's own language codes.
- [Model requirements](https://docs.clika.io/modelverse/model-requirements.md): Memory and device figures per model variant, for sizing a deployment before anything downloads.
- [Supported models](https://docs.clika.io/modelverse/supported-models.md): The models the release serves, read off its supported-models document: one row per model with its weight variants and the backends the release's run record covers.

## Platform

- [Platform](https://docs.clika.io/platform.md): The CLIKA Platform runs benchmarks on real devices, keeps a fleet of them under management, serves models from them, and issues the licenses ClikaRT needs.
- [Using the CLI](https://docs.clika.io/platform/cli.md): clika-cli, the CLIKA Platform command-line client: install it, log in, and drive every platform operation from a terminal or a CI job.
- [apply and export](https://docs.clika.io/platform/cli/apply-export.md): clika-cli apply and export: the resource YAML round trip that keeps a definition in git identical to the definition the platform runs.
- [Artifacts and models](https://docs.clika.io/platform/cli/artifacts.md): clika-cli artifacts and models: upload files the platform pushes to devices, register models from Hugging Face, and manage versions and retention.
- [Authentication and profiles](https://docs.clika.io/platform/cli/authentication.md): clika-cli login, logout and profiles: the API key the CLI signs in with, where it is stored, and how to address several deployments from one machine.
- [Benchmarks](https://docs.clika.io/platform/cli/benchmarks.md): clika-cli benchmarks and benchmark-groups: launch a benchmark across devices, watch it to completion, and read the measured results.
- [Cheat sheet](https://docs.clika.io/platform/cli/cheatsheet.md): The twenty clika-cli commands worth memorising, in one table.
- [Cloud instances](https://docs.clika.io/platform/cli/cloud.md): clika-cli cloud-instances, cloud-credentials, cloud-images and vpn: renting a machine in a cloud account and enrolling it as a device.
- [Devices](https://docs.clika.io/platform/cli/devices.md): clika-cli devices: find the machines registered with your platform, run commands on them, move files, and manage their desired state.
- [Get started with the CLI](https://docs.clika.io/platform/cli/get-started.md): Ten minutes from nothing to your first clika-cli commands: install the CLI from your deployment, log in, look around, and read a benchmark run from the terminal.
- [Job definitions](https://docs.clika.io/platform/cli/job-definitions.md): clika-cli job-definitions: the reusable description of a piece of work the platform runs on a device.
- [Jobs](https://docs.clika.io/platform/cli/jobs.md): clika-cli jobs: dispatch work to a device, follow it, and pull back its logs, outputs and per-sample results.
- [Licensing](https://docs.clika.io/platform/cli/licensing.md): clika-cli license, projects licenses and entitlement-profiles: activating a deployment, and issuing the credentials a runtime uses.
- [MCP server and generic dispatch](https://docs.clika.io/platform/cli/mcp.md): clika-cli mcp, tools and call: expose the platform to an AI assistant as Model Context Protocol tools, or dispatch any API operation by name.
- [Model deployment](https://docs.clika.io/platform/cli/model-deployment.md): clika-cli servings: deploy a model behind an endpoint on a device, and call it through the platform.
- [Events, metrics and alerts](https://docs.clika.io/platform/cli/monitoring.md): clika-cli events, metrics, alerts, notifications, audit and webhooks: watching what the platform is doing and being told when it changes.
- [Other resource groups](https://docs.clika.io/platform/cli/other-resources.md): The remaining clika-cli resource groups, and a complete index of every top-level command with the page that documents it.
- [Organizations, projects and access](https://docs.clika.io/platform/cli/projects-and-orgs.md): clika-cli orgs, projects, users, roles and permissions: the tenancy and access-control commands.
- [Keeping the CLI current](https://docs.clika.io/platform/cli/self-update.md): clika-cli self-update and version: pull a newer CLI from the deployment's own release channel, with checksum and signature verification.
- [Services and service definitions](https://docs.clika.io/platform/cli/services.md): clika-cli service-definitions and services: the long-running processes the platform keeps alive on a device.
- [Transfers](https://docs.clika.io/platform/cli/transfers.md): clika-cli transfers: follow a file transfer between the platform and a device to completion, and cancel one that is no longer wanted.
- [Concepts](https://docs.clika.io/platform/concepts.md): The objects the platform is built from, what each one is for, and how they connect.
- [Artifact](https://docs.clika.io/platform/concepts/artifact.md): The versioned files the platform stores and moves onto devices: uploads, directories, external Docker and Git sources, tags, deduplication and retention.
- [Benchmark](https://docs.clika.io/platform/concepts/benchmark.md): What a benchmark run is, how the platform turns it into work on your devices, and every benchmark in the catalogue: what it is, what it tests, and how it is scored.
- [Device](https://docs.clika.io/platform/concepts/device.md): A machine running the CLIKA agent: how it registers, what it reports, the states it moves through, and what blocks it from taking work.
- [Job](https://docs.clika.io/platform/concepts/job.md): One execution of one benchmark on one device: the definition it runs, the states it moves through, and what it guarantees on the way out.
- [Model deployment](https://docs.clika.io/platform/concepts/model-deployment.md): Running one of your models on your own devices as an endpoint: the platform proxy, the endpoints a served model exposes, the deployment page, and how ClikaRT does the work.
- [Organization and project](https://docs.clika.io/platform/concepts/organization-and-project.md): The two containers every resource sits in: the organization that owns users and devices, and the project that scopes one piece of work.
- [Roles](https://docs.clika.io/platform/concepts/roles.md): The organization roles and the project roles as the platform enforces them: what each may do, where each is set, and the points still being clarified.
- [Runtime licenses](https://docs.clika.io/platform/concepts/runtime-licenses.md): The credential a ClikaRT runtime presents to prove it may run: the two kinds, every entitlement it can carry, entitlement profiles, and the lifecycle.
- [Service](https://docs.clika.io/platform/concepts/service.md): A process the platform keeps running on a device: its definition, desired state and reconciliation, and how it differs from a model deployment.
- [First steps](https://docs.clika.io/platform/getting-started.md): New to the CLIKA Platform? Start here. What it is, the five-minute mental model, and a tutorial that ends with a benchmark result you can read.
- [Add a model](https://docs.clika.io/platform/getting-started/first-benchmark/add-a-model.md): Register a Hugging Face model with the platform, and understand why registration moves no weights and why the task matters.
- [Add your first device](https://docs.clika.io/platform/getting-started/first-benchmark/add-your-first-device.md): Register a machine with the CLIKA agent: the install command, the activation code it carries, and what each of Linux, macOS, Windows and Android does differently.
- [Overview](https://docs.clika.io/platform/getting-started/first-benchmark/overview.md): A tour of the signed-in platform: what each section of the navigation holds, and how the organization and project switchers decide what you see.
- [Pick a device](https://docs.clika.io/platform/getting-started/first-benchmark/pick-a-device.md): Find your device in the fleet, read what the platform knows about it, and learn the device page tab by tab.
- [Read the results](https://docs.clika.io/platform/getting-started/first-benchmark/read-the-results.md): Read a finished benchmark run: the headline cards, the quality and performance sections, the per-sample evidence behind a score, and how to share it.
- [Run a benchmark](https://docs.clika.io/platform/getting-started/first-benchmark/run-a-benchmark.md): Create a benchmark run: pick the devices and the models, then the quality tests, launch it, and watch the platform fan it out into jobs.
- [What to read next](https://docs.clika.io/platform/getting-started/next-steps.md): Where the Platform documentation goes after your first benchmark: concepts, how-to guides, the CLI and the MCP surface.
- [The Platform at a glance](https://docs.clika.io/platform/getting-started/overview.md): What the CLIKA Platform is and is not, and the five-minute mental model behind every page of these docs.
- [Quickstart: CLI](https://docs.clika.io/platform/getting-started/quickstarts/cli.md): Install clika-cli, sign in with an API key, register a device, run and read a benchmark, and deploy, from a terminal.
- [Quickstart: licensing](https://docs.clika.io/platform/getting-started/quickstarts/licensing.md): Issue a runtime license in a project, download ClikaRT from the platform, and run a licensed prompt on your machine in six steps.
- [Quickstart: MCP](https://docs.clika.io/platform/getting-started/quickstarts/mcp.md): Give Claude Code the platform's benchmark tools with a scoped key, and run the first benchmark from a prompt.
- [Quickstart: web](https://docs.clika.io/platform/getting-started/quickstarts/web.md): From a fresh account to a benchmark result and a running endpoint in the browser, with the minutes each step takes.
- [Sign up with a signup code](https://docs.clika.io/platform/getting-started/sign-up-with-a-signup-code.md): Create an organization and its owner account from a signup link or code, verify the address, and know what the plan behind the code allows.
- [How-to guides](https://docs.clika.io/platform/how-to.md): Problem-oriented recipes for the CLIKA Platform, grouped by what you are trying to get done.
- [Activate a deployment license](https://docs.clika.io/platform/how-to/activate-a-deployment-license.md): Activate a platform license on a fresh deployment, keep it renewed, and issue the runtime credentials a ClikaRT project needs.
- [Download the ClikaRT SDK](https://docs.clika.io/platform/how-to/download-the-clikart-sdk.md): Get the ClikaRT runtime, the clikart-cli model CLI and the C++ SDK for your own machine from a licensed platform deployment: the web dialog, the platform CLI and the MCP tools, and how to verify the archive.
- [Manage members and API keys](https://docs.clika.io/platform/how-to/manage-members-and-api-keys.md): Add people to an organization and a project, give them the right role, and mint a scoped API key for an integration.
- [Read benchmark results](https://docs.clika.io/platform/how-to/read-benchmark-results.md): Pull a benchmark run's numbers and its per-sample evidence: what each metric means, the API routes behind the result view, and how to diagnose a leg that failed.
- [Register a device by code](https://docs.clika.io/platform/how-to/register-a-device-by-code.md): Enroll a device with a short activation code: the code the Register Device dialog's command carries, and how to mint one from the CLI or an MCP client.
- [Update an on-premise platform](https://docs.clika.io/platform/how-to/update-an-on-premise-platform.md): Learn when a newer platform release is published, apply it with the installer, and return to the previous version from the backup the upgrade takes.
- [Write a job definition](https://docs.clika.io/platform/how-to/write-a-job-definition.md): The job definition YAML: every key explained, a complete worked example, the environment the platform injects, and how to apply the file to a deployment.
- [Write a service definition](https://docs.clika.io/platform/how-to/write-a-service-definition.md): The service definition YAML: every key explained, a worked example, how a definition becomes a running service, and what is not shipped yet.
- [Using with AI (MCP)](https://docs.clika.io/platform/mcp.md): Connect Claude or ChatGPT to the CLIKA Platform and ask about your devices and benchmarks in plain words: which page to follow for your app, the API key it needs, what it can and cannot do, and the endpoint details for advanced users.
- [ChatGPT Desktop](https://docs.clika.io/platform/mcp/chatgpt-desktop.md): Add the CLIKA Platform to ChatGPT Desktop from its Settings, check that it works, and remove it.
- [Claude Code](https://docs.clika.io/platform/mcp/claude-code.md): Connect Claude Code to the platform with clika-cli mcp install, with claude mcp add, or with an entry in ~/.claude.json, then check the connection and remove it.
- [Claude Desktop](https://docs.clika.io/platform/mcp/claude-desktop.md): Add the CLIKA Platform to Claude Desktop with the extension from the platform's Settings page, check that it works, and remove it.
- [Codex](https://docs.clika.io/platform/mcp/codex.md): Connect the Codex CLI and the Codex IDE extension to the platform through config.toml, with clika-cli mcp install, with codex mcp add, or by hand, then check the connection and remove it.
- [Other clients](https://docs.clika.io/platform/mcp/other-clients.md): Connect any MCP client that speaks Streamable HTTP to the platform with the endpoint URL and an Authorization header, or run the CLIKA CLI as a local stdio server for a client that only starts commands.
- [Troubleshooting](https://docs.clika.io/platform/mcp/troubleshooting.md): What to do when an MCP client is refused with a 401, cannot verify a self-signed certificate, or does not show the platform's tools after a change.