Liner Projects and Mixtapes

Liner saves each job as a durable Liner Project. A Mixtape is the curated research packet inside that project.

Project Foldermobile-design-foundations/

liner.yaml

LINER.md

SKILL.md

working/

mixtape/

MIXTAPE.md

tape.yaml

synthesis.md

sources/

01-touch-target-size.md

02-apple-gestures.md

local-sources/

touch-target-paper.pdf

working/

01-jtbd-and-knowledge-map.md

02-candidate-longlist.md

03-evaluation.yaml

04-quality-checks.md

Tape

mixtape/tape.yaml is the recipe. It names the mixtape, describes its purpose, identifies the curator, and lists the kept sources.

The tape is small enough to review, fork, and store in Git.

For supported changes after creation, use Liner’s Project and Source maintenance commands instead of editing canonical files directly. Core preserves identity, plans the exact file effects, and records a Change Receipt after apply.

Durable Identity And Maintenance Records

Each current Project has an immutable Project ID in liner.yaml. Each canonical Source has an immutable Source ID in mixtape/tape.yaml. Display names, folder paths, URLs, titles, notes, and sections can change without becoming identity.

• Project rename and move preserve the Project ID.
• Source metadata and locator updates preserve the Source ID.
• Semantic Source replacement creates a successor Source ID and records predecessor lineage.
• Source removal retains content and provenance; purge is a separate destructive action.

Before a supported change, Liner reads a versioned Project Snapshot and creates a write-free Project Change Set for that exact revision. After apply, it writes a redacted Change Receipt under mixtape/.liner-runs/maintenance/. The receipt records identity, effects, validation, stale derived artifacts, and next actions without copying Source bodies or private notes.

Liner Project

The Liner Project owns the Mixtape plus the instruction files future AI sessions need.

• LINER.md tells an AI how to use the saved research.
• SKILL.md is the default skill file for local AI assistants.
• liner.yaml records status, evidence pointers, and instruction-file state.
• working/ is for audits and maintenance notes.

Mixtape

The Mixtape is the research packet. It lives under mixtape/ and includes the tape, synthesis, extracted sources, optional private source copies, methodology notes, and compiled output.

Consumable Output

mixtape/MIXTAPE.md is the research entry point. It includes the curator synthesis and a source index with links to extracted files in mixtape/sources/. A written MIXTAPE.md means the research packet is compiled, but the AI instruction files may still need to be created.

Instruction Files

In the terminal UI, Create Operating Layer is the guided action that writes LINER.md, root SKILL.md, and liner.yaml.

Use it when a future AI session should load the whole folder, follow the method from the saved research, and understand status without reading every working note first.

Curator Notes

Every source should explain why it belongs. A source with no note is just a bookmark. A useful note tells the AI what matters, what to skim, and any limitation to keep in mind.

Skills And Projects

Skills carry reusable method: steps, style, rules, and taste.

Mixtapes carry job-specific research context: primary docs, examples, contested takes, and local files.

A Liner Project combines both: job-specific evidence in the Mixtape, plus task-specific method in the root skill file.