aidp-migrator-bootstrap

# `aidp-migrator-bootstrap` — environment readiness check Confirms everything the migrator needs is in place before any of the other skills attempt real work. Idempotent — re-runnable whenever you suspect drift. ## When to use - The user is running the migrator for the first time on this workstation. - Any other skill fails with an OCI auth, network, or cluster-not-found error. - The user moved to a new region / DataLake / workspace and you need to re-verify. ## Required env-coords Before any check below works, the user MUST have an `env-coords.md` (or any plain notes file) listing **all of these**. Refuse to proceed and ask for them if any are missing — never guess. | Coordinate | Example shape (do NOT use any of these literal values — these are placeholders) | |---|---| | AIDP REST base URL | `https://aidp.<region>.oci.oraclecloud.com/20240831` | | DataLake OCID | `ocid1.aidataplatform.oc1.<region>.<id>` | | Workspace UUID | `<8-4-4-4-12>` UUID format | | Cluster ID | `<8-4-4-4-12>` UUID format | | OCI profile name | `<your-profile>` (e.g. the section name in `~/.oci/config`) | | Output workspace path | `Shared/aidp-migration-tool-output/` (or your team's path) | Save these into a `env-coords.md` file at the project root, gitignored. Every other skill in this plugin threads these through verbatim. See [references/env-coords.md](../../references/env-coords.md) for a complete scaffold. ## Step-by-step ### 1. Python prereqs ```bash python3 --version # 3.10+ python -m pip install -r ~/.aidp-migrator/engine/requirements.txt ``` Expected packages: `oci`, `requests`, `websocket-client`, `openai`, `cryptography`. If any is missing, install + retry. ### 2. OpenAI env vars ```bash python3 -c "import os; print('OPENAI_API_KEY=' + ('set' if os.getenv('OPENAI_API_KEY') else 'missing'))" python3 -c "import os; print('OPENAI_MODEL=' + os.getenv('OPENAI_MODEL', '<team-approved-default>'))" ``` If `OPENAI_API_KEY` is empty, the migrator's Pass-2 cell-by-cell loop will not run. Ask the user to set `OPENAI_API_KEY` in the shell. If the toolkit requires an explicit model, ask them to set `OPENAI_MODEL` to the team's approved OpenAI model. ### 3. OCI authentication Two valid paths — confirm which the user is using: **API key (recommended for unattended runs)** ```bash oci iam region list --profile <profile-name> # returns a region list on success ``` **Session token (interactive only, expires ~1 hr)** ```bash oci session validate --profile <profile-name> # if expired: oci session authenticate --profile <profile-name> --region <region> ``` If either returns 401/403 or "Security Token expired", surface the exact command the user should run — do not pretend success. ### 4. AIDP cluster reachability + state ```bash curl -s -H "..." \ "<AIDP_BASE>/dataLakes/<DATALAKE_OCID>/workspaces/<WORKSPACE_UUID>/clusters/<CLUSTER_ID>" \ | python3 -c "import json,sys; d=json.load(sys.stdin); print(d.get('lifecycleState'))" ``` Expect `Active`. Acceptable transient states are `Starting`, `Updating`. If `Stopped` or `Failed`, instruct the user to start the cluster from AIDP console before any Pass-2 run. > Curl with `auth=signer` (the migrator's helper) is the proper way; if the user wants an exact one-liner, suggest it in Python (fill the four placeholder values from your `env-coords.md`): > ```python > import oci, requests > > # Fill these from env-coords.md (see references/env-coords.md): > profile = "<your-profile-name>" > base = "<AIDP_BASE>" # e.g. https://aidp.<region>.oci.oraclecloud.com/20240831 > lake = "<DATALAKE_OCID>" > ws = "<WORKSPACE_UUID>" > cl = "<CLUSTER_ID>" > > cfg = oci.config.from_file(profile_name=profile) > signer = oci.signer.Signer(cfg["tenancy"], cfg["user"], cfg["fingerprint"], cfg["key_file"]) > r = requests.get(f"{base}/dataLakes/{lake}/workspaces/{ws}/clusters/{cl}", auth=signer) > print(r.json().get("lifecycleState")) > ``` ### 5. Workspace write permission Confirm the operator can write under the target output path: ```bash python3 -c " import oci, requests, urllib.parse, os cfg = oci.config.from_file(profile_name='<profile>') signer = oci.signer.Signer(cfg['tenancy'], cfg['user'], cfg['fingerprint'], cfg['key_file']) base = '<AIDP_BASE>' lake = '<DATALAKE_OCID>' ws = '<WORKSPACE_UUID>' path = '<output-workspace-path>' r = requests.get(f'{base}/dataLakes/{lake}/workspaces/{ws}/objects', params={'path': path}, auth=signer) print(r.status_code, r.text[:200]) " ``` 200 → good. 404 → ask the user to create the folder via AIDP console or via `POST .../objects` with `type=FOLDER`. 401/403 → the operator's OCI principal lacks `workspace.write` on this workspace. ### 6. Migrator pyc compile sanity ```bash python3 -m py_compile \ ~/.aidp-migrator/engine/scripts/build_dag.py \ ~/.aidp-migrator/engine/scripts/check_data_availability.py \ ~/.aidp-migrator/engine/scripts/job_migrate.py \ ~/.aidp-migrator/engine/scripts/migrate_catalog.py ``` Should be silent. If a Python-version mismatch is reported, the user is on the wrong interpreter. ## Output template After running, report back to the user as a single table: ``` | Check | Status | Notes | |---|---|---| | Python deps | OK | 3.13.x, all 5 deps present | | OPENAI_API_KEY | OK | set in shell env | | OPENAI_MODEL | OK | team-approved model, if required | | OCI auth (api_key) | OK | profile <name>, region <region> | | Cluster state | OK | Active | | Workspace write | OK | output base reachable | | Migrator compile | OK | all 4 entrypoints compile | ``` Any FAIL row → emit the exact command the user needs to run to fix it. Do not proceed to other skills if any check failed. ## Notes - This skill never modifies anything. It is read-only. - If the user is on a session token and it's about to expire (`oci session validate` shows <10 minutes remaining), warn them and suggest refresh BEFORE invoking a long-running skill like [`aidp-migrate-job`](../aidp-migrate-job/SKILL.md). - Re-run after any change in region, DataLake, workspace, or cluster.