See feature guide if you’re wondering how something specific here works.

• 
  • a green eye means that the link was visited before and has some contexts, associated with it. When you click on it, the sidebar opens with more information about the visits.
  • You can see that I’ve sent the link to Jane on Telegram (#1)
  • I’ve annotated the link on Instapaper and highlights (#3) is marked with yellow right within the page

    Normally in order to see see your Instapaper highligths you have to go to the web app first!

  • I’ve clipped some content to my personal notes at some point (#8), the selected text was matched and highlighted as well

    If I click notes/ip2org.org, it will cause my Emacs to open and jump to the file, containing this note.

• 
  • I have a note about the page in my personal notes on my filesystem (#2)
  • I chatted with a friend and sent them the link at some point (#6)

    Clicking on the link will open Telegram and jump straight to the message where the link was mentioned. So you can reply to it without having to search or scroll over the whole chat history.

  • I’ve tweeted about this link before (#11)

    Similarly, clicking would jump straight to my tweet.

  • I also have this link annotated via Hypothesis (#4, #12, #13)
• Annotated demo of sidebar with direct & child visits, and ‘visited’ marks (click to zoom)

• Annotated demo of highlight and locator features (click to zoom)

• More:
  • A video, demonstrating
    • ‘mark visited’ feature
    • highlights
    • jumping to the originating tweet; jumping straight to the org-mode heading where links occurred
  • You can find more short screencasts, demonstrating various features here
  • There are more screenshots here
  • This post features a video demo of using Promnesia with a Roam Research database
  • Video demos for the mobile extension:
    • showing all references to pages on W3C website in my org-mode notes
    • marking already visited links on the page, to make easier to process Reddit posts

• browser extension
  • neatly displays the history and other information in a sidebar
  • handles highlights
  • provides search interface

  However, browser addons can’t read access your filesystem, so to load the data we need a helper component:

• server/backend: promnesia serve command

  It’s called ‘server’, but really it’s just a regular program with the only purpose to serve the data to the browser. It runs locally and you don’t have to expose it to the outside.

• indexer: promnesia index command

  Indexer goes through the sources (specified in the config), processes raw data and extracts URLs along with other useful information.

  Another important thing it’s doing is normalising URLs to establish equivalence and strip off garbage. I write about the motivation for it in “URLs are broken”.

You might also want to skim through the glossary if you want to understand deeper what information Promnesia is extracting.

Promnesia ships with some builtin sources. It supports:

• data exports from online services: Reddit/Twitter/Hackernews/Telegram/Messenger/Hypothesis/Pocket/Instapaper, etc.

  It heavily benefits from HPI package to access the data.

• Google Takeout/Activity backups
• Markdown/org-mode/HTML or any other plaintext on your disk
  • for ex. local personal knowledge bases - Logseq (example config), Obsidian
• in general, anything that can be parsed in some way
• you can also add your own custom sources, Promnesia is extensible

See SOURCES for more information.

Here’s a diagram, which would hopefully help to understand how data flows through Promnesia.

See HPI section on data flow for more information on HPI modules and data flow.

Also check out my infrastructure map, which is more detailed!

┌─────────────────────────────────┐ ┌────────────────────────────┐ ┌─────────────────┐
│ 💾       HPI sources            │ │  💾    plaintext files      │ │  other sources  │
│ (twitter, reddit, pocket, etc.) │ │ (org-mode, markdown, etc.) │ │ (user-defined)  │
└─────────────────────────────────┘ └────────────────────────────┘ └─────────────────┘
                                ⇘⇘              ⇓⇓               ⇙⇙
                                 ⇘⇘             ⇓⇓              ⇙⇙
                                 ┌──────────────────────────────┐
                                 │ 🔄    promnesia indexer      │
                                 |        (runs regularly)      │
                                 └──────────────────────────────┘
                                                ⇓⇓
                                 ┌──────────────────────────────┐
                                 │ 💾    visits database        │
                                 │       (promnesia.sqlite)     │
                                 └──────────────────────────────┘
                                                ⇓⇓
                                 ┌──────────────────────────────┐
                                 │ 🔗    promnesia server       │
                                 |       (always running)       |
                                 └──────────────────────────────┘
                                                ⇣⇣
                                 ┌─────────────────────────────────┐
         ┌───────────────────────┤  🌐      web browser            ├────────────────────┐
         │  💾 browser bookmarks ⇒      (promnesia extension)      ⇐  💾 browser history |
         └───────────────────────┴─────────────────────────────────┴────────────────────┘

• extension:
  • Chrome: desktop version. Unfortunately mobile Chrome doesn’t support web extensions.
  • Firefox: desktop and Android
    • note: web extensions on Android are mostly broken at the moment, see here (unless you’re using Firefox Nightly)
    • unfortunately, iOS Firefox doesn’t support web extensions.
  • you can also find ‘unpacked’ versions in Releases

    It can be useful because Chrome Web Store releases might take days to approve, but in general the store version if preferrable.

• backend

  Note that Promnesia can work without the backend, so technically this step is optional. But then it will only be able to use browser visits and browser bookmarks, so the benefits of the extension will be limited.

  • simplest: install from PyPi: pip3 install --user promnesia
    • install optional dependencies with: pip3 install --user bs4 lxml mistletoe logzero
  • alternatively: you can clone this repository and run it as scripts/promnesia

    This is mainly useful for tinkering with the code and writing new modules.

  You might also need some extra dependencies. See “Extra dependencies” for more info.

  As for supported operating systems:

  • Linux: everything is expected to work as it’s what I’m using!
  • OSX: expected to work, but there might be issues at times (I don’t have any macs so working blind here). Appreciate help!

    You might want to run brew install libmagic for proper MIME type detection.

  • Windows: at the moment doesn’t work straightaway (I don’t have any Windows to test against), there is an open issue describing some workarounds.
  • Android: allegedly, possible to run with Termux! But I haven’t got to try personally.

You can try out Promnesia with the extension only, it uses browser history and bookmarks as data sources. However, Promnesia’s main power is using data from external sources, and for that you’ll need to run the indexer and backend (more on it in the next section).

The easiest way to try this mode is to run promnesia demo command, it can give you a sense of what Promnesia is doing with zero configuration.

1. Install the extension and the server (PIP package), in case you haven’t already
2. Run promnesia demo https://github.com/karlicoss/exobrain

   This clones the repository, (my personal wiki in this case), extracts the URLs, and runs on the port 13131 (default, can be specified via --port)

   You can also use a path on your local filesystem (e.g. directory with text files), or a website URL.

3. After that, visit https://www.reddit.com

   If you press the extension icon, you will see the pages from my blog where I link to posts on Reddit.

To get the most benefit from Promnesia, it’s best to properly setup your own config, describing the sources you want it to use. If something is unclear, please feel free to open issues or reach me, I’m working on improving the documentation. Also check out troubleshooting guide or open discussions on documentation.

• create the config: promnesia config create

  The command will put a stub promnesia config in your user config directory, e.g. ~/.config/promnesia/config.py on Linux. (it’s possibly different on OSX and Windows, see this if you’re not sure). If you wish to specify a custom location, you can set the PROMNESIA_CONFIG environment variable or pass the --config flag.

• edit the config and add some sources

  You can look at an example config, or borrow bits from an annotated configuration example here: doc/config.py.

  The only required setting is:

  • SOURCES

    SOURCES specifies the list of data sources, that will be processed and indexed by Promnesia.

    You can find the list of available sources with more documentation on each of them here: SOURCES.

    • reading example config: doc/config.py
    • browsing the code: promnesia/sources.

  If you want to learn about other settings, the best way at the moment (apart from reading the source) is, once again, example config.

  • [optional] check the config

    First, you can run promnesia doctor config, it can be used to quickly troubleshoot typos and similar errors. Note that you may need to install [mypy](https://github.com/python/mypy) for some config checks.

    Next, you can use the demo mode: promnesia demo --config /path/to/config.py.

    This will index the data and launch the server immediately, so you can check that everything works as expected in your browser.

• run the indexer: promnesia index

  At the moment, indexing is periodic, not realtime. The best is to run it via cron/systemd once or several times a day:

  # run every hour in cron
  0 * * * *      promnesia index    >/tmp/promnesia-index.log 2>/tmp/promnesisa-index.err
      

  Note: you can also pass --config /path/to/config.py explicitly if you prefer or want to experiment.

  • [optional] check the database

    Run promnesia doctor database to quickly inspect the database and check that stuff that you wanted got indexed. You might need to install sqlitebrowser first.

• run the server: promnesia serve

  You only have to start it once, it will automatically detect further changes done by promnesia index.

  • [optional] autostart the server with promnesia install-server

    This sets it up to autostart and run in the background:

    • via Systemd for Linux
    • via Launchd for OSX. I don’t have a Mac nearby, so if you have any issues with it, please report them!

    I think you can also use cron with @reboot attribute:

    # sleep is just in case cron starts up too early. Prefer systemd script if possible!
    @reboot     sleep 60 && promnesia serve   >/tmp/promnesia-serve.log 2>/tmp/promnesia-serve.err
            

    Alternatively, you can just create a manual autostart entry in your desktop environment.

  • [optional] check that the server is responding promnesia doctor server
• [optional] setup MIME handler to jump to files straight from the extension

  See a short 20s demo, and if this is something you’d like, follow the instructions in open-in-editor.

Visit represents an ‘occurence’ of a link in your digital trace. Obviously, visiting pages in your browser results in visits, but in Promnesia this notion also captures links that you interacted with in other applications and services.

In code (python, JS), visits are reprented as class Visit (and class DbVisit).

Visits have the following fields:

• url: hopefully, no explanation needed!

  The only required field.

• timestamp: when the page was visited

  Required, but in the future might be optional (sometimes you don’t have a meaningful timestamp).

• locator: what’s the origin of the visit?

  Usually it’s a permalink back to the original source of the visit.

  For example:

  • locators for a link extracted from Reddit data point straight into reddit.com interface, for the corresponding post or comment
  • locators for a link extracted a local file point straight into these files on your disk. Clicking on the locator will open your text editor via MIME integration

  Required, but in the future might be optional. (TODO also rename to ‘origin’??)

• context: what was the context, in which the visit occurred?

  For example:

  • context for Telegram visits is the message body along with its sender
  • context for a link from org-mode file is the whole paragraph (outline), in which it occurred

  I usually call a visit without a context ‘boring’ – it doesn’t contain much information except for the mere fact of visiting the page before. However they are still useful to have, since they fill in the gaps and provide means of tracing through your history.

  Optional.

• duration: how much we have spent on the page

  Somewhat experimental field, at the moment it’s only set for Chrome (and often not very precise).

  Optional.

Digression: now that you have an idea what is a Visit, you can understand few more things about Promnesia:

• source (or indexer) is any function that extract visits from raw files and generates a stream of visits (i.e. Iterable[Visit]).
• promnesia indexer goes through the sources, specified in config, collects the visits and puts in the database
• promnesia server reads visits form the database, and sends them to the extension

Now let’s consider some concrete examples of different kinds of Visits:

• Google Takeout indexer

  Results in visits with:

  • url
  • timestamp
  • locator

  There isn’t any context for visits from takeout, because it’s basically a fancy database export.

• Instapaper indexer

  Generates a visit for each highlight on the page:

  • url: original URL of the annotated page
  • timestamp: time when you created the highlight
  • locator: permalink to the highlight, bringing you into the Instapaper web app
  • context: highlight body
• Markdown indexer

  Extracts any links it finds in Markdown files:

  • url: extracted link
  • timestamp: Markdown doesn’t have a well defined datetime format, so it’s just set to the file modification time.

    However, if you do have your own format, it’s possible to write your own indexer to properly take them into the account.

  • locator: links straight into the markdown file on your disk!
  • context: the markdown paragraph, containing the url

Note: this terminology is not set is stone, so if someone feels there are words that describe these concepts better, I’m open to suggestions!

See doc/GUIDE.org#FAQ

The best support for me would be if you contribute to this or my other projects. Code, ideas of feedback – everything is appreciated.

I don’t need money, but I understand it’s often easier to give away than time, so here are some of projects that I donate to:

• org-mode
• Internet Archive
• Hypothes.is
• Doom Emacs

• Troubleshooting guide
• Documentation on the sources
• Developer’s guide
• Privacy policy