Machine learning · 2026
Agent Memory Workflow
Pick up the work without repeating the whole history.
Long projects need continuity. This workflow keeps decisions and useful lessons available when a session ends or work moves to another machine. It combines an existing open-source vault with a portable record of the practice’s own project context.
- Status
- Practice tooling
- Domain
- Developer workflow · Persistent context
- Role
- Workflow configuration and cross-machine setup
- Year
- 2026
Explore the workflow with illustrative content.
Keep the original input
The first stage represents an incoming record. A separate source copy makes it possible to inspect where an extracted value began.
Check the structured result
Fields are extracted into a consistent shape. Switch to the review example to see a missing field routed for attention.
Send it to the right next step
A complete sample reaches the output. A sample with a missing field remains in review rather than silently passing through.
Illustrative content · no live data
Give the next session a useful starting point
Stack
- Markdown
- Git
- Obsidian
- Agent hooks
A working setup for carrying project decisions between coding sessions and machines, using plain-text records, an Obsidian vault and a separate portable memory repository.
Build on existing work
The vault uses Obsidian Mind, the open-source project by Brenno Ferrari.
The work described here is its use within the practice and the surrounding memory workflow; authorship of the upstream tool belongs to its maintainers.
Keep a portable record
A separate ilang-memory repository holds a cross-machine record in plain Markdown.
Git makes that context inspectable and portable, while project-local notes can capture details close to the code.
Know which copy is current
Session hooks can create notes inside individual projects, which makes it easy for copies to drift.
The setup names a canonical cross-machine record and treats local histories as separate working context, keeping that distinction visible.
Pipeline
A configured memory workflow
Stage Can reject
-
01
Read context
Relevant project notes
-
02
Work
Decisions and lessons
-
03
Record
Plain-text context
Upstream tools plus project notes
-
04
Review
Current shared record
-
05
Carry forward
Context for the next session
Behind the work
The skills
that came together.
Workflow
- Cross-machine context management
- Plain-text decision records
- Agent tool configuration
- Memory ownership and drift management
Project highlights
-
Context that travels
Project decisions remain available across machines
-
Plain-text records
Readable and versioned without a hosted service
-
Upstream credited
Uses Obsidian Mind by Brenno Ferrari