ResearchFlow

approved

by Nachiket Tanksale

A Markdown-first research and project operating system for Obsidian. Connect projects, tasks, daily work, ideas, reading, and career opportunities. - This plugin has not been manually reviewed by Obsidian staff.

↓ 127 downloadsUpdated 1mo agoApache-2.0

Install in Obsidian View on GitHub

ResearchFlow

Install ResearchFlow from the Community Plugins

A Markdown-first research and project operating system connecting ideas, projects, tasks, daily work, reading, and career opportunities.

ResearchFlow is a Markdown-first research and project operating system for Obsidian.

It connects:

Research ideas
ML and Quantum work
Projects
Tasks
Daily work
Reading
Career opportunities
Blockers
Attention items
Decisions
Results and artifacts

The central idea is simple:

Connect what you think about, what you build, what you read, what you work on, and where you want to go next.

Current Status

Version: 0.9.2

ResearchFlow is an actively developed alpha. The current implementation combines the original foundation with the v0.2--v1.0 roadmap features.

The plugin is intentionally local and Markdown-first. Your vault remains usable as normal Obsidian Markdown even without the plugin.

What ResearchFlow Does

ResearchFlow treats your research workflow as a connected graph rather than a collection of unrelated folders.

                    ResearchFlow
                         │
       ┌─────────────────┼─────────────────┐
       │                 │                 │
      Ideas           Projects           Career
       │                 │                 │
       │               Tasks               │
       │                 │                 │
       └────────────── Daily ───────────────┘
                         │
                  Reading / Results

A typical workflow can look like:

Reading
   ↓
Research Idea
   ↓
Project
   ↓
Task
   ↓
Daily Work
   ↓
Result / Artifact
   ↓
Career Opportunity

The relationships are represented using normal Markdown, YAML frontmatter, and Obsidian wikilinks.

Core Features

1. Central Dashboard

The ResearchFlow home view provides a single landing page for your work.

It currently shows:

Active projects
Project progress
Project priority
Project deadlines
Project health
Blockers
Attention flags
Stale projects
Today's incomplete tasks
Upcoming career opportunities
Reading queue
Research ideas
Project timeline / health information
Overall counts and status statistics

The dashboard is intended to answer:

What is happening, what is blocked, and what needs my attention?

2. Projects

Projects are the primary unit of active work.

A project stores:

Domain
Project type
Status
Priority
Progress
Start date
Deadline
Next action
Blocker
Attention flag
Last activity
Milestones
Tasks
Decisions
Artifacts
Related reading
Related ideas
Related career opportunities
Daily work

Example:

---
type: project
domain: ML
status: active
priority: high
progress: 45
start: 2026-08-01
deadline: 2026-09-15
blocker:
attention: false
next_action: Run the baseline experiment
last_activity: 2026-08-18T20:30:00
project_kind: research
---

Project progress

Project progress is derived from task completion.

Completed tasks
---------------- × 100
All project tasks

The task state is therefore the source of truth rather than manually maintaining a percentage.

3. Tasks

Tasks can be simple work items while still having their own Markdown page.

A task can contain:

Objective
Architecture
Code links
Tests
Artifacts
Issues
Decisions
Results
Daily work

Example:

---
type: task
status: todo
priority: medium
project: "[[Astronomy Agent]]"
created: 2026-08-18
work_date: 2026-08-18
due:
---

Project relationship

Every task can belong to a project.

When a task is created:

Task
 ├──► Project
 └──► Daily Note

The project task list is maintained automatically.

4. Daily Work

Daily notes are the common record of what actually happened during the day.

A single day can contain work on multiple projects:

2026-08-18

OpportunityAgent
Astronomy Agent
Ising Research

The daily note contains:

Today's Focus
Tasks
Work Log
Decisions
Blockers
Ideas
Reading
Career

Tasks scheduled for a date are automatically represented in the corresponding daily note.

Daily ↔ Task Synchronization

Task completion is synchronized across views.

             ┌──────────────┐
             │     TASK     │
             │  status=todo │
             └──────┬───────┘
                    │
        ┌───────────┼───────────┐
        ▼           ▼           ▼
     Landing      Daily       Project
        │           │           │
        └───────────┴───────────┘
                    │
                    ▼
              status = done

Changing a task in the daily workflow updates the underlying task object and consequently updates project progress and dashboard state.

Completed tasks are removed from the dashboard's active daily task list.

5. Research Ideas

Ideas are captured before they become projects.

ResearchFlow distinguishes between:

Research ideas
Project ideas

and supports domains such as:

ML
Quantum
General

An idea can contain:

Hypothesis
Motivation
Related work
Possible experiments
Open questions
Next action
Related projects
Related reading
Notes

Example:

---
type: idea
domain: Quantum
kind: research
status: seed
priority: medium
created: 2026-08-18
project:
---

6. Reading

Reading items are structured objects rather than an unstructured bookmark dump.

A reading item tracks:

Title
URL
Type
Status
Date added
Date read
Notes
Takeaways
Related project
Related ideas
Why it was saved

Example:

---
type: reading
reading_type: paper
status: unread
added: 2026-08-18
read:
url: https://example.com/paper
project: "[[Astronomy Agent]]"
---

The dashboard provides a reading queue.

7. Career

Career opportunities are stored as Markdown objects and surfaced in the ResearchFlow dashboard.

A career opportunity can track:

Company
Role
Deadline
Match percentage
Status
Application date
Feedback
Documents
Related project
Source URL
Daily work

Example:

---
type: career
company: Example AI
role: ML Researcher
deadline: 2026-09-01
match: 87
status: saved
applied:
feedback:
documents:
project: "[[LLM Research]]"
source: https://example.com/job
---

Career opportunities are sorted by deadline in the dashboard.

OpportunityAgent Integration

ResearchFlow supports importing career opportunities from a CSV export.

The intended workflow is:

OpportunityAgent
       │
       ▼
    CSV export
       │
       ▼
ResearchFlow Career objects
       │
       ▼
Career dashboard

CSV import is deliberately used as the initial integration boundary rather than requiring a direct dependency between the two projects.

8. Project Health

ResearchFlow tracks project health using available project metadata and activity.

Health takes into account factors such as:

Project progress
Priority
Blockers
Attention flags
Deadlines
Recent activity
Staleness

Projects without recent activity are surfaced as stale.

The current default stale threshold is 14 days.

9. Blockers and Attention

Projects can explicitly declare blockers:

blocker: Waiting for dataset access

and attention:

attention: true

The dashboard separates these into:

🔴 BLOCKERS

🟠 ATTENTION

⚠ STALE PROJECTS

This makes unresolved work visible without requiring manual searching through project notes.

10. Weekly Research Summary

ResearchFlow can generate a weekly Markdown summary.

The summary provides a place for:

Weekly activity
Wins
Blockers
Decisions
Next-week planning

The generated summary remains a normal Markdown file inside the vault.

11. Vault Validation

ResearchFlow includes a relationship validation command.

It checks for issues such as:

Tasks pointing to missing projects
Tasks missing work_date
Invalid project relationships

Use:

Command Palette → ResearchFlow: Validate ResearchFlow Relationships

Commands

ResearchFlow currently provides commands for:

Command Purpose

Open Home Open the ResearchFlow dashboard

New Project Create a project

New Research Idea Create an idea

New Task Create a task

Open Today's Daily Note Open/synchronize today's daily note

New Reading Create a reading item

New Career Opportunity Create a career opportunity

Import Career CSV Import OpportunityAgent-style career data

Generate Weekly Research Summary Generate a weekly research summary

Validate ResearchFlow Relationships Check relationship consistency

Folder Structure

The default structure is:

02_Projects/
03_Ideas/
04_Tasks/
05_Career/
06_Reading/
07_Daily/

These folders can be changed from the ResearchFlow settings tab.

ResearchFlow creates the folders automatically if they do not exist.

Data Model

ResearchFlow uses Markdown files with YAML frontmatter.

Project
  │
  ├── Tasks
  │     └── work_date → Daily
  │
  ├── Reading
  │
  ├── Ideas
  │
  └── Career

Daily
  ├── Tasks
  ├── Work Log
  ├── Decisions
  ├── Blockers
  ├── Ideas
  ├── Reading
  └── Career

The important design principle is:

Objects are stored once; relationships are represented through links and metadata.

Design Principles

Markdown First

ResearchFlow does not require a proprietary database.

Your information remains in:

Markdown files
YAML frontmatter
Obsidian wikilinks

If ResearchFlow disappears tomorrow, your vault remains usable.

Links Over Duplication

ResearchFlow prefers relationships over duplicated information.

For example:

project: "[[Astronomy Agent]]"

allows the same task to appear conceptually in:

the task file
the project
the daily note
the dashboard

without creating separate copies of the task.

Task State as Source of Truth

For task completion:

Task file
   │
   ├──► Daily view
   ├──► Project progress
   └──► Dashboard

The task's status is authoritative.

Project progress is derived from project tasks rather than being an independent manually maintained number.

Local First

The core plugin does not require:

A server
An account
A database
A cloud service
An API key

All core data lives in your Obsidian vault.

Roadmap

The version roadmap describes the evolution of ResearchFlow.

v0.1 --- Foundation

v0.2 --- Synchronization

v0.3 --- Research Context

ML / Quantum / General categorization
Research / Project idea types
Idea → Project relationship
Reading → Project relationship
Career → Project relationship
Project milestones
Project activity tracking
Project timeline information

v0.4 --- Career

v0.5 --- Intelligence

v1.0 --- Integrated Research Operating System

Future beyond v1.0

Potential future capabilities include:

Knowledge graph visualization
Research/project analytics
LLM-assisted linking
Automatic daily summaries
Project retrospectives
Career/project feedback analysis
Research idea discovery
Deeper OpportunityAgent integration
XLSX import
More advanced workload planning

Development

ResearchFlow is built using:

TypeScript
Obsidian Plugin API
npm
TypeScript compiler
esbuild

Install dependencies:

npm install

Development build:

npm run dev

Production build:

npm run build

The production build generates:

main.js

inside the plugin directory.

For development, use a separate Obsidian test vault rather than your primary vault.

Project Structure

obsidian-research-flow/
│
├── manifest.json
├── package.json
├── tsconfig.json
├── esbuild.config.mjs
├── README.md
├── LICENSE
│
└── src/
    └── main.ts

Installation for Development

Clone the repository.
Run:

npm install

Build:

npm run build

Copy the plugin directory into:

<Vault>/.obsidian/plugins/obsidian-research-flow/

Make sure the directory contains:

manifest.json
main.js
styles.css

Enable ResearchFlow from Obsidian's Community Plugins settings.

Current Limitations

ResearchFlow is still an alpha release.

Known limitations include:

CSV is the current career import format; direct XLSX import is not included.
The plugin does not currently depend on an external database.
Some advanced analytics and graph views remain future work.
The exact appearance of the dashboard depends on the accompanying styles.css.
Career and reading workflows are functional but intentionally lightweight compared with a dedicated database application.
Automatic intelligence features are rule-based rather than LLM-driven.

Philosophy

ResearchFlow is not intended to become another generic task manager.

The goal is to make the following loop easy:

Think
  ↓
Capture an idea
  ↓
Explore / read
  ↓
Turn it into a project
  ↓
Break it into tasks
  ↓
Work on it daily
  ↓
Record decisions and results
  ↓
Connect the outcome to future work

The system should help answer four questions quickly:

What am I working on?
What needs attention?
What did I actually accomplish?
What should I work on next?

License

Apache License 2.0.

For plugin developers

Search results and similarity scores are powered by semantic analysis of your plugin's README. If your plugin isn't appearing for searches you'd expect, try updating your README to clearly describe your plugin's purpose, features, and use cases.