michellegdyason173 downloadsA plugin for deep self-discovery journaling using AI. Use analysis history, and export options. Requires an OpenAI API key.
AI-powered journaling, multi-perspective analysis, goal generation, and reflective chat for Obsidian.
There's more to life than what psychiatry knows about.
Active repository: github.com/MichelleGDyason/The_Obsidian_Deleometer_v2
The Deleometer is a reflective conversation and journaling tool. It is not a medical device, diagnosis, treatment, or substitute for medication, therapy, crisis support, or professional care already used by the journal author.
AI can make mistakes and can sound more certain than it is. Treat Deleometer responses as invitations to think with, question, revise, and discuss, not as truth to absorb undiluted. Use it like a rigorous conversation with a thoughtful friend, not an all-knowing or infallible robot. Continuing to chat with the AI can help mistakes become clearer, because you can challenge the response, ask for evidence, request a simpler explanation, or ask it to reinterpret the same entry through another frame.
If you use OpenAI mode, your OpenAI API key is hidden in the settings screen, but it is still stored locally in Obsidian plugin data. If your vault, device backups, or plugin data are synced or shared, treat the key as sensitive.
The Deleometer is an Obsidian plugin, not a sealed or encrypted vault. AI features send the relevant journal, chat, goal, author memory, and analysis context to the AI provider selected in settings. OpenAI mode sends context to OpenAI. Local Ollama mode sends context only to the configured local endpoint and does not silently fall back to OpenAI. The settings include privacy controls for redacting common sensitive details before AI calls, keeping author memory local, excluding the personality profile from prompts, clearing stored author memory, clearing the API key, avoiding full-journal context in perspective chat, and checking the local Ollama model library for changes.
Saved analyses, chats, goals, author memory, and plugin settings are stored locally as Obsidian data or Markdown. The Deleometer does not encrypt those files. Anyone with access to the vault, backups, sync provider, device account, or another sufficiently powerful Obsidian plugin may be able to read them. Install the plugin only from the MichelleGDyason GitHub repository or the official Obsidian community plugin listing when available, and treat BRAT beta releases as development builds.
File discovery is limited to the journal, goals, milestones, and Full Calendar folders configured in The Deleometer settings. The plugin does not enumerate every Markdown file in the vault. Author memory is updated incrementally from journal entries the user explicitly analyses; it does not scan the journal folder or the rest of the vault to learn about the author.
This project is licensed under the GNU Affero General Public License v3.0.
Why AGPL-3.0:
As copyright holder, Michelle G Dyason may choose to relicense future versions under MIT if community adoption clearly requires a more permissive license. That does not change the license of versions already released under AGPL.
The Deleometer helps you:
The plugin keeps an author memory summary so later analyses can respond as part of an ongoing reflective conversation with previous analyses. This is meant to help recurring themes, values, supports, risks, and patterns become easier to notice over time.
1.7.2 or laterObsidian42 - BRAT is the easiest way to install and update beta builds of The Deleometer before it is available in Obsidian's official community plugin directory.
Settings -> Community plugins.Obsidian42 - BRAT, and enable it.BRAT: Add a beta plugin for testinghttps://github.com/MichelleGDyason/The_Obsidian_Deleometer_v2
or:
MichelleGDyason/The_Obsidian_Deleometer_v2Latest unless you intentionally want to freeze testing on one specific beta release.Add Plugin and wait for BRAT to finish downloading the plugin.Settings -> Community plugins, refresh the installed plugin list if needed, find The Deleometer, and enable it.Settings -> Community plugins -> The Deleometer and complete the setup steps below.To update a BRAT install later, open the command palette and run:
BRAT: Check for updates to all beta plugins and UPDATE
If The Deleometer does not appear after BRAT installs it, restart Obsidian, then check that your vault contains .obsidian/plugins/the-deleometer/ with manifest.json, main.js, and styles.css inside it. If BRAT reports an error, remove the beta plugin entry from BRAT and add the repository again.
.obsidian/plugins/the-deleometer/manifest.jsonmain.jsstyles.cssThe Deleometer in Community Plugins.npm install
npm run build
main.jsmanifest.jsonstyles.cssObsidian does not load main.ts directly. It must be built into main.js.
After enabling the plugin:
Settings -> Community plugins -> The Deleometer.Full Calendar Folder to the folder watched by your Full Calendar local source.Local Ollama models are local files. They do not come bundled with The Deleometer and they do not update themselves inside The Deleometer. The user must download at least one model in Ollama before local AI can work. In settings, use Find installed models to detect the models already available on the machine, then choose one from the installed model dropdown. If no models are found, use Download model into Ollama in settings, or install one directly in Ollama. Starter examples include llama3.1, qwen2.5, or mistral.
The Deleometer records the selected model digest and local model library signature, and shows a notice when your local Ollama model library changes. Use Find installed models again after pulling or replacing local models.
The settings screen also shows an approximate cost estimate based on how many analysis lenses are selected. OpenAI mode estimates API spend using gpt-4o-mini token pricing. Local Ollama mode has no OpenAI/API cost, but the user pays in local model download size, disk space, battery, CPU/GPU use, and slower runtime depending on the computer and model.
Save & Analyze.Analyses can be extensive. With all 261 frames enabled, an appended note may produce around 62,500 words, roughly a 300-minute read. Settings provide family shortcuts and one chronological catalogue for choosing individual analyses, so you can tune the output to the time and attention you have. Generated analyses follow historical chronology first. Period and intellectual-family labels provide context but never move a perspective out of its temporal milieu.
The analysis reader level can also be changed in settings. The available levels range from Grade 5 primary through professor-level analysis. This changes how much the AI explains specialist terms and how far it tries to guide the reader beyond their current zone of proximal development. The output language can also be changed in settings for generated analyses, syntheses, goals, chat replies, and journaling prompts. Current language options are English, French, and German.
These captures show the current Obsidian workflow: long-form analysis output, group synthesis and goal sections, the journal-entry composer, and the reflective assessments modal.




When I began my undergraduate course at university, I was learning new theories about self and subjectivity and wondered what it would be like if those theories could work as therapy. The Deleometer grew out of that question.
I am not a coder, but with the help of OpenAI's Codex I have been able to build a form of theoretical therapy intended for anyone who wants to think with more than one framework at once. The plugin treats a journal entry as something that can be read through philosophy, psychoanalysis, social theory, religion, cultural studies, and other traditions, not only as mood tracking or productivity data.
The goal is reflective agency. The author is meant to argue with the AI, learn from it, reject parts of it, and use the conversation to make their own meanings clearer.
The plugin can create dated goal and milestone event notes that the Full Calendar plugin can display.
It supports:
Sync goals to Full CalendarThe plugin can turn the milestones inside each goal into first-class Markdown notes under Deleometer/Milestones by default. This makes milestones easier to review, search, link, and reorganise in Obsidian.
Milestone support includes:
Sync milestones to folderConsolidate similar milestonesConsolidation keeps one merged milestone note and marks the duplicate milestone notes as merged redirects, so the history of where the milestone came from is not simply erased.
The plugin includes commands for:
Before publishing widely, you will probably want to:
package.jsonversions.json in sync with each releaseIf you are testing a BRAT prerelease, please open a GitHub issue and include the exact beta version, what you expected, what happened instead, and the steps that led there.
Even a short note saying "BRAT beta X works for me on platform Y" is useful, because it helps show whether anyone besides the maintainer is actively testing.
You may also want to add:
This project is licensed under the GNU Affero General Public License v3.0. See LICENSE.
The Deleometer currently includes 261 individual analysis frames across 11 intellectual families. Generated analyses are ordered by chronology, beginning with accumulated Australian Indigenous philosophy and then moving through ancient, classical, modern, and contemporary traditions so the analyses can show a progression in thought. A perspective's family is descriptive metadata rather than a shelf: it remains in its given period. Family syntheses are produced separately after the chronological individual analyses have been explicated.