servant-docs-simple: Generate endpoints overview for Servant API

[ library, mit, utility ] [ Propose Tags ]

This library uses Data.Typeable to generate documentation for Servant API types. It relies on the typeRep of Servant's combinators and other datatypes used in the API to generate the documentation.

[Skip to Readme]
Versions [RSS] [faq],,,
Change log
Dependencies aeson (>=1.4.2 && <1.5), aeson-pretty (>=0.8.7 && <0.9), base (>= && <4.14), bytestring (>= && <0.11), ordered-containers (>=0.2.2 && <0.3), prettyprinter (>=1.2.1 && <1.7), servant (>=0.15 && <0.18), text (>= && <1.3), unordered-containers (>=0.2.10 && <0.3) [details]
License MIT
Copyright 2020 Holmusk
Author kwannoel
Maintainer Holmusk <>
Category Utility
Home page
Bug tracker
Source repo head: git clone
Uploaded by HolmuskTechTeam at 2020-10-10T10:11:39Z
Distributions NixOS:
Downloads 307 total (17 in the last 30 days)
Rating (no votes yet) [estimated by Bayesian average]
Your Rating
  • λ
  • λ
  • λ
Status Hackage Matrix CI
Docs available [build log]
Last success reported on 2020-10-10 [all 1 reports]


[Index] [Quick Jump]


Maintainer's Corner

For package maintainers and hackage trustees


Readme for servant-docs-simple-

[back to package description]


Build status Hackage Stackage Lts Stackage Nightly MIT license


This library uses Data.Typeable to generate documentation for Servant API types.

It relies on the typeRep of Servant's combinators and other datatypes used in the API to generate the documentation.

Why do we need this?

  1. We need the API Format types inside the documentation. They can be used as keys to look up examples, fields, and other miscellaneous details.

  2. We want to generate an overview of our endpoints from the information our API type provides, without having to write instances.

In-depth explanation here


Example usage

Using this script

-- stack --system-ghc runghc --package servant-docs-simple
{-# LANGUAGE DataKinds        #-}
{-# LANGUAGE TypeApplications #-}
{-# LANGUAGE TypeOperators    #-}

module Main where

import Servant.API ((:>), Post, ReqBody)
import Servant.Docs.Simple (writeDocsJson, rriteDocsPlainText)

-- Our API type
type API = "hello" :> "world" :> Request :> Response
type Request = ReqBody '[()] ()
type Response = Post '[()] ()

main :: IO ()
main = do
  -- Writes to the file $PWD/docsJson
  writeDocsJson @API "docs.json"

  -- Writes to the file $PWD/docsPlainText
  writeDocsPlainText @API "docs.txt"

Expected Output

Files should be generated relative to $PWD

$ ls | grep docs


    "/hello/world": {
        "Response": {
            "Format": "': * () ('[] *)",
            "ContentType": "()"
        "RequestType": "'POST",
        "RequestBody": {
            "Format": "': * () ('[] *)",
            "ContentType": "()"


    Format: ': * () ('[] *)
    ContentType: ()
RequestType: 'POST
    Format: ': * () ('[] *)
    ContentType: ()


Click on these links for tutorials

Generating plaintext/JSON documentation from api types

Generating the intermediate documentation structure

Writing our own rendering format

Parsing custom API combinators

To run tutorial scripts

# clone this repository
git clone

# run the examples
stack examples/<source file>


What's the usecase for this?

  1. Lightweight documentation solution by providing an overview of endpoints.

  2. Keeping documentation for types in one place

    Suppose we have a library of format types. These are shared between different languages and formats. If we centralize all the information in this library, we have one source of truth.

    --                                           --- Format Types ---
    --                                          |                    |
    --                                          v                    v
    type singleAPI = "hello" :> ReqBody '[JSON] User :> POST '[JSON] Message
    data User = User Name Password deriving Typeable 
    data Messsage = Message Id Content deriving Typeable

    We use these format types in our API as illustrated above. Developers can use these format types to index the library, to find information (examples, instances, ...) they need about the type.

What is required to achieve this?

  1. We need documentation to be generated from the API type, without writing extra instances for each format type we use.

    We can do this with typeRep from Data.Typeable.

    $ stack ghci
    Prelude> import Data.Typeable
    Prelude Data.Typeable> data Alignment = Good | Bad deriving Typeable
    Prelude Data.Typeable> typeRep (Proxy :: Proxy Alignment)
  2. We need access to the API format types (Email, Users, ...) in documentation.

    This is so we can index our library with these types to get the relevant fields, examples for these.

How does our usecase differ from Servant.Docs?

  1. Servant.Docs generates documentation only if Format Types have implemented the necessary instances.

    -- Instances for format types; they provide examples for format types
    instance ToSample User where
        toSamples _ = <some example>
    instance ToSample Message where
        toSamples _ = <some example>

    Hence, we cannot generate documentation solely from the Format Type.

  2. In documentation generated by Servant.Docs, the format types mentioned above (User, Message) are not included. This means we can't use them to index our library to look for the relevant information.

What are the trade-offs?

  1. Currently Servant.Docs.Simple does not support as many formats as Servant.Docs (via Servant.Docs.Pandoc).

  2. Examples are not mandatory as you do not have to write instances for your format types.

    This means that if developers don't provide examples, the generated documentation would be without them.

    We can still provide these examples through:

    • The shared library of types

    • Using Servant combinators such as Summary and Description. These are generated as part of Documentation through Servant.Docs.Simple.