> For the complete documentation index, see [llms.txt](https://docs.openg2p.org/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.openg2p.org/tools/g2p-advisor/scripts.md).

# Scripts

Utility scripts under `scripts/` in the [advisor repo](https://github.com/openg2p/g2p-advisor). All are TypeScript, run via `tsx`, and pick up environment from `.env`.

## `npm run migrate`

Apply every SQL file under `migrations/` in lexical order, idempotently. Each migration runs inside a transaction. The `schema_migrations` table tracks which files have been applied so re-runs are safe.

Run after pulling new commits if the repo's `migrations/` directory has new files.

## `npm run seed:phase2`

Stuff a known-good Phase 2 input set into a project's `working_case`. Bypasses the (planned) Phase 2 chat handler — use it during development to test the build pipeline without going through the conversational input-collection flow.

```bash
npm run seed:phase2 -- \
  --project <project-uuid> \
  --gitlab-handle <your-gitlab-username> \
  --gitlab-email <your-email> \
  [--mnemonic health-worker] \
  [--org-mnemonic openg2p] \
  [--advance-phase] \
  [--dry-run]
```

Seeds: a single Register called "Health Workers Register" with 5 simple columns, an `HW-` Functional ID prefix, sandbox + production hostnames, `small` Helm resource profile, your GitLab credentials. Clears Phase 1 conditional triggers (`notification_triggering_events`, `existing_system_to_replace`) so Phase 2's conditional checks don't fire.

`--mnemonic` drives the registry's repo + Python package names. `--org-mnemonic` is the implementing organisation's stamp on every Docker image name (`<org>-<mnemonic>-<service>:develop`).

## `npm run cleanup:phase2`

Full Phase 2 cleanup for one project. Leaves Phase 1 state intact.

```bash
npm run cleanup:phase2 -- --project <project-uuid> \
  [--keep-gitlab-repos] [--keep-workspace] [--keep-current-phase] \
  [--clear-messages] [--dry-run]
```

What it does (in order):

1. `TRUNCATE build_jobs + build_job_events` — kills every build job + event row in the DB.
2. Strips every Phase 2 `working_case` key from this project; rolls `current_phase` back to 2 (Phase 1 stays approved).
3. *(optional, `--clear-messages`)* — `DELETE FROM project_messages WHERE project_id = …`. Wipes the chat history for this project so stale LLM hallucinations don't get replayed into the next conversation. Captured Phase 1 facts in `working_case` are NOT touched — only the message thread.
4. `rm -rf <WORKSPACE_PATH>/<project-dir>/` — clears the local generated trees and reference clone.
5. `DELETE` the project's GitLab repos: `<mnemonic>-extension` and `<mnemonic>-deployment`.

What it doesn't touch: the project row itself, Phase 1 working\_case keys, approved Phase 1 reports, the user's GitLab subgroup, other projects' workspaces. `project_messages` only deleted with `--clear-messages`.

`--dry-run` previews everything without applying.

## `npm run db:up` / `npm run db:down`

Start / stop the local Postgres container defined at `docker/docker-compose.yml`. Used during dev so you don't need a local Postgres install.


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:

```
GET https://docs.openg2p.org/tools/g2p-advisor/scripts.md?ask=<question>&goal=<endgoal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.