Automattic / Automattic/data-liberation-agent
Restore standalone Data Liberation Agent as source authority
- Dominant language
- TypeScript
- Stars
- 31
- Forks
- 3
- Avg merge
- 10h 14m
- Merged PRs (30d)
- 81
Description
## Objective
Restore this repository as the authoritative home for Data Liberation Agent and reduce Studio to a thin consumer plus its generic local-site orchestration.
## Current state
- The standalone repository stopped receiving changes on June 30, 2026.
- The package moved to `Automattic/studio/packages/data-liberation-agent` on July 1.
- Studio PR Automattic/studio#3952 now contains both generic `studio create --from` orchestration and substantial DLA capture/artifact work.
- This coupling has made the Studio branch large, conflict-prone, and difficult to review independently.
## Ownership boundary
Data Liberation Agent owns:
- source detection and platform adapters
- browser acquisition, hydration, screenshots, and provenance
- portable website artifact generation
- `/liberate` skills, MCP tools, and agent bundles
- theme-replication and blocks/product routing
Studio owns:
- `studio create --from`
- local site lifecycle and Blueprint/native PHP execution
- Static Site Importer installation, continuation, cleanup, and Studio UX
The integration contract is a versioned portable website artifact consumed through Studio’s generic `studio create --from` boundary.
## Migration plan
1. Extract `packages/data-liberation-agent` history from the current Studio candidate and merge it into this repository without discarding either repository’s history.
2. Re-establish standalone install, build, lint, test, bundle-drift, and release contracts.
3. Port the DLA-owned changes from Studio PR #3952 here.
4. Publish an immutable standalone DLA release/package.
5. Change Studio to consume that immutable release through a thin integration.
6. Remove the duplicate in-tree package only after parity and end-to-end verification.
7. Reduce Studio PR #3952 to Studio-owned orchestration and tests.
## Acceptance criteria
- One authoritative DLA source tree exists in this repository.
- Claude Code, Codex, and direct CLI/MCP entry points work from the standalone package.
- Studio consumes a pinned immutable DLA version rather than maintaining a forked source tree.
- Portable website artifacts pass schema, boundedness, provenance, and deterministic reconstruction checks.
- Studio’s generic import path accepts the artifact and completes an end-to-end local WordPress import.
- No DLA implementation remains duplicated in Studio after cutover.
- Automattic/studio#3952 is reduced to its generic Studio lifecycle boundary.
## AI assistance
GPT-5.6 Sol via OpenCode audited the standalone repository, the current Studio package, and Studio PR #3952; it helped define the ownership boundary and migration acceptance criteria. Chris Huber reviewed and orchestrated the work.
Contributor guide
Research direction
Start with packages/data-liberation-agent and compare the standalone repository with the DLA work in Studio PR #3952. Review the standalone install, build, lint, test, bundle-drift, and release contracts, then trace the CLI and MCP entry points. Done means the standalone package is authoritative, its artifacts pass the listed checks, and Studio consumes a pinned immutable release without duplicated implementation.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- typescript
- Domain
- backend, cli, tooling
- Issue type
- Refactor
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Active
- Clarity
- Mostly clear
- Newbie friendliness
- 25/100