Migration Companion

contentstack-migration-companion

contentstack-migration-companion

Guide a user through migrating a project from Contentful to Contentstack — first the content (content types, entries, assets, locales) via the Contentstack CLI migrate plugin, then the website code that reads from the CMS.

This is a sequential workflow where each step produces output that the next step consumes (the create command produces a populated stack and a bundle with credentials, which feeds the code migration). Treat the artifact paths and counts that each command prints as state you must capture and carry forward.

Operating principles

Follow these throughout — they matter more than any single command:

• Work in a unique session workspace. At the very start of Step 1, create a session-scoped directory by running:

  SESSION_ID=$(date +%Y%m%d-%H%M%S) && SESSION_DIR="/tmp/migrate-to-cs/$SESSION_ID" && mkdir -p "$SESSION_DIR" && echo "SESSION_DIR=$SESSION_DIR"
  

  Record the printed SESSION_DIR value (e.g. /tmp/migrate-to-cs/20260608-143022) and carry it as a concrete literal through every shell command in this migration — do not regenerate it. This keeps each migration run isolated so concurrent sessions and re-runs never collide. If the user points you to a different workspace, use their path instead.
• Bundled scripts & references live next to this skill — resolve them via {SKILL_DIR}. This skill ships helper scripts (a scripts/ folder) and reference docs (a references/ folder) alongside this SKILL.md file. Wherever these instructions write {SKILL_DIR}, substitute the absolute path of the directory this SKILL.md was loaded from — i.e. the skill's own install directory. Do not assume a fixed path. The location differs by AI assistant, by OS, and by whether the skill was installed per-project or per-user — for example it may be <project>/.claude/skills/contentstack-migration-companion, ~/.claude/skills/contentstack-migration-companion, or a Windows path like %USERPROFILE%\.claude\skills\contentstack-migration-companion. Determine the real path once (it is the folder you read this SKILL.md from; if unsure, search the workspace and home directory for */contentstack-migration-companion/SKILL.md), and for shell commands set it as a variable up front (SKILL_DIR="<that absolute path>") so bundled scripts can be invoked as "$SKILL_DIR/scripts/<name>". The bundled scripts self-locate their own siblings, so once you invoke them by absolute path they work regardless of your current directory.
• Pin the Node version for the whole session. A machine often has several Node versions (system, Homebrew, multiple nvm installs) and a non-interactive shell may resolve an old one (e.g. /usr/local/bin/node v14) ahead of the user's nvm default. The Step 1 prereq checker finds the highest installed Node ≥ 20 and reports its directory as node.bin_dir in the JSON. Record that value as a concrete literal NODE_BIN_DIR and prefix every csdx, npm, and contentful command for the rest of the migration with it, e.g. PATH="<NODE_BIN_DIR>:$PATH" csdx migrate:create …. This guarantees the CLI runs on the Node the prereq check validated, not whatever an unconfigured shell picks first. If node.bin_dir is absent (older check output), fall back to the plain command.
• One step at a time, and show the result. After each command, surface the meaningful output to the user — the summary tables, the counts, the artifact path — not a wall of raw logs. The user is watching this like a progress bar; give them a clean status, then the path/handle the next step needs.
• Track migration progress with the checklist. At each [PROGRESS] trigger below, output a progress block in your response using these emoji: ✅ = completed, ⏳ = currently running, ⬜ = not yet started. Example for Step 3 in progress:

  **Migration progress**
  - ✅ Step 1 — Prerequisites & inputs
  - ✅ Step 2 — Install migrate plugin
  - ⏳ Step 3 — Content Migration
  - ⬜ Step 4 — Code Migration
  

  Output it at two moments: (1) at the start of each step, and (2) when each step's eval passes. Only one step is ⏳ at a time. Step 4 stays ⏳ through all sub-steps 4.1–4.6. Step 5 (Welcome) is not tracked — it triggers automatically once Step 4 is ✅.
• Gate the destructive or expensive steps. Confirm before logging in, before creating a new stack, and before editing the user's code. These either touch live accounts or modify their repo, so a quick "ready to proceed?" prevents nasty surprises.
• Capture the outputs explicitly. When a command prints a bundle path, a log directory, or a stack API key, record the exact path/value and reuse it verbatim. Do not guess paths — read them back from the command output.
• Browser-based login is normal here. csdx auth:login --oauth opens the user's browser; the terminal then blocks and auto-detects when they finish. Run it, tell the user to complete login in the browser, and simply wait for the command to return — do not try to script the browser or kill the command.
• Currently Contentful is the only supported source. Do not ask the user which legacy platform they're on; assume Contentful. (The CLI flag is --source contentful.)
• If a step fails, stop and diagnose rather than barrelling ahead. Most failures here are recoverable (expired token → re-login, missing content model → inform the user), and the relevant recovery is described in the step that can fail.
• Never display code in your text output. Do not show shell commands, code snippets, scripts, or any fenced code blocks (``` blocks) in your chat messages at any point during the migration. Just run commands silently and report the result in plain prose. The user sees tool calls in the tool panel — repeating code in chat is noise.

The migration at a glance

Work through them in order. The sections below give the exact commands, what to show the user, and what to carry forward.

Self-contained skill. Everything Step 4 needs — the full code-migration procedure, the Contentful → Contentstack SDK reference (references/), and the eval scripts (scripts/) — ships inside this one skill. There is no separate code-migration skill to install; resolve the bundled files via {SKILL_DIR} as described above.

Step 1 — Prerequisites & inputs

[PROGRESS] Output the migration progress block: Step 1 → "in_progress", Steps 2–4 → "pending".

1.0 — Create session workspace

Before doing anything else, create the unique session directory for this migration run:

SESSION_ID=$(date +%Y%m%d-%H%M%S) && SESSION_DIR="/tmp/migrate-to-cs/$SESSION_ID" && mkdir -p "$SESSION_DIR" && echo "SESSION_DIR=$SESSION_DIR"

Record the printed SESSION_DIR path exactly (e.g. /tmp/migrate-to-cs/20260608-143022). Substitute this concrete value wherever these instructions reference $SESSION_DIR. Do not regenerate it — every step in this migration must use the same directory so artifacts chain correctly.

Tell the user: "Session workspace created at <SESSION_DIR>."

1.1 — Detect the Python 3 command

Run this to find the correct Python 3 command for this environment:

if python3 --version 2>&1 | grep -q "Python 3"; then
  PYTHON_CMD=python3
elif python --version 2>&1 | grep -q "Python 3"; then
  PYTHON_CMD=python
else
  PYTHON_CMD=""
fi
echo "PYTHON_CMD=$PYTHON_CMD"

Record PYTHON_CMD exactly as printed. Substitute $PYTHON_CMD wherever these instructions show a Python invocation — do not hardcode python3 or python.

If PYTHON_CMD is empty, stop immediately and tell the user:

"Python 3 is required but not found. Install it from python.org or via your package manager (e.g. brew install python3 on macOS, sudo apt install python3 on Ubuntu, or download the installer from python.org on Windows), then try again."

Do not proceed past this point until Python 3 is detected.

1.2 — Run the prerequisite checker

Run this single script. It silently evaluates Node.js, installs any missing CLIs (csdx, contentful), checks the Contentstack region and login, and checks the Contentful login and spaces — all in one pass. It outputs a JSON summary:

$PYTHON_CMD "{SKILL_DIR}/scripts/check_prereqs.py"

Parse the JSON result and carry every field forward as session state.

Record the Node bin directory. Capture node.bin_dir from the JSON as the concrete literal NODE_BIN_DIR. Per the "Pin the Node version" principle in the overview, prefix every later csdx, npm, and contentful command with PATH="<NODE_BIN_DIR>:$PATH" so the whole migration runs on the Node the checker validated — not whatever an unconfigured non-interactive shell would resolve first (the checker already picks the highest installed Node ≥ 20, scanning PATH and all nvm installs). If node.bin_dir is missing, fall back to the plain command.

Hard blocker: If the script exits with code 1, Node.js is missing or too old. The reported node.version/node.path reflect the best Node found anywhere on the machine, so the message is accurate even when a newer Node exists but isn't on the default PATH. Stop immediately and tell the user the exact problem:

• node.error == "not_installed" → "Node.js is not installed. Install it via nvm install 22 or from nodejs.org, then try again."
• node.ok == false (e.g. node.version == "v18.x") → "The newest Node I can find is <version>, but Node 20+ is required. Install a newer one with nvm install 22 && nvm use 22 (or upgrade your system Node), then try again."

Do not continue past this point until Node 20+ is confirmed.

1.3 — Handle missing Contentstack login (if needed)

Skip this sub-step if cs_login.ok is true in the JSON.

If cs_login.ok is false (needs_login) or cs_login.org_uid is null (needs_oauth_reauth), trigger a fresh OAuth login:

csdx auth:login --oauth

Tell the user: "A browser window is opening — complete the Contentstack login there, then come back here." Wait for the command to return (it blocks until the browser flow finishes), then re-run the prereq checker (same command as 1.2) to capture the updated email and org UID.

If the org UID is still missing after the retry, tell the user:

"OAuth login did not store an org UID. Please re-run csdx auth:login --oauth manually, then let me know when done so I can retry."

1.4 — Handle missing Contentful login (if needed)

Skip this sub-step if contentful_login.ok is true in the JSON.

contentful login is interactive (it needs the user to press Y and paste a token) — you cannot run it as a Bash command. Instruct the user to run it themselves:

"Please run this command in your terminal:

contentful login

It will ask 'Continue login on the browser? (Y/n)' — press Y. A browser window will open — sign in there. When the browser login completes, the terminal will show 'Paste your token here:' — copy your Management Token from the browser page and paste it, then press Enter. Come back here once the login confirms success."

Wait for the user to confirm they have completed the login, then re-run the prereq checker to pick up the new session.

The Management Token is a secret: do not ask the user to share it with you, do not echo it back in your summaries, and do not write it into any file in this workspace.

1.5 — Show the prerequisites summary and confirm

Once cs_login.ok and contentful_login.ok are both true, display a summary table from the JSON result. Use ✅ for items that look good and ⚠️ for anything that may need attention:

Then ask this single question:

"Everything looks good — ready to proceed? Or would you like to change anything before we start?

• proceed — start the migration
• region — switch the Contentstack region
• contentstack login — switch the Contentstack account
• contentful login — switch the Contentful account"

Wait for the user's answer before continuing.

Region picker — when the user asks to change the region, show this menu and wait for their choice before running the command:

"Which region is your destination stack in?

1. AWS-NA — AWS North America
2. AWS-EU — AWS Europe
3. AWS-AU — AWS Australia
4. AZURE-NA — Azure North America
5. AZURE-EU — Azure Europe
6. GCP-NA — Google Cloud North America
7. GCP-EU — Google Cloud Europe"

Map the number to its region code and run:

csdx config:set:region <REGION>   # e.g. csdx config:set:region AWS-EU

Repeat the summary + question until the user confirms they are ready to proceed.

1.6 — Contentful Space ID (select from list)

The contentful_spaces array in the prereq JSON already holds all accessible spaces (populated from the contentful space list call run inside the checker).

If there is exactly one space in the list, select it automatically — do not ask the user. Announce the selection:

"Found one Contentful space: My Marketing Site (abc123). Using it automatically."

Capture its ID as SPACE_ID and continue.

If there are two or more spaces, present them as a numbered menu — do not ask the user to type or paste a Space ID:

1. My Marketing Site  (abc123)
2. Developer Sandbox  (def456)
…

Ask:

"Which space do you want to migrate? Enter the number."

Wait for the answer. Map the number back to the Space ID and capture it as SPACE_ID.

If the list is empty (no spaces parsed), fall back to asking:

"I couldn't list your Contentful spaces. Please enter the Space ID directly."

Then confirm the token can reach the selected space:

contentful space use --space-id <SPACE_ID>

If this fails, the token likely lacks access to that space — tell the user and ask them to re-check the selection or switch to a Contentful account that has the right access.

Eval — Verify Step 1 before proceeding

# Use the Node the checker validated (NODE_BIN_DIR from the prereq JSON).
export PATH="<NODE_BIN_DIR>:$PATH"
$PYTHON_CMD --version   # must print Python 3.x
node --version          # must print v20.x or higher
csdx --version          # must print a version number
csdx auth:whoami        # must print a logged-in email
contentful --version    # must print a version number
contentful space list   # must return a list of spaces — not "You have to be logged in"

(export only affects this one Eval block — shell state does not persist across commands, so later steps must still prefix PATH="<NODE_BIN_DIR>:$PATH" per the overview principle.)

Pass criteria:

• $PYTHON_CMD detected and exits 0 with Python 3.x
• Node major version ≥ 20
• csdx auth:whoami returns an email (not "No user logged in")
• cs_login.org_uid is non-null (captured from prereq JSON in step 1.2 / 1.3)
• contentful space list returns rows without an auth error
• SPACE_ID captured from the user's selection in 1.6

Note: contentful whoami is not available in all CLI versions — use contentful space list to verify authentication instead.

If every check passes:

[PROGRESS] Output the migration progress block: Step 1 → "completed", Step 2 → "in_progress", Steps 3–4 → "pending".

Then proceed to Step 2. If any check fails, fix the issue (re-run the relevant sub-step) and re-verify.

Step 2 — Install the migrate CLI plugin

[PROGRESS] Output the migration progress block: Step 1 → "completed", Step 2 → "in_progress", Steps 3–4 → "pending".

The csdx migrate:* commands come from @contentstack/cli-external-migrate. Run this block to install or update it automatically — it is intentionally silent unless something changes or fails:

PLUGIN_NAME="@contentstack/cli-external-migrate"

# Version currently installed (empty string if not installed)
INSTALLED=$(csdx plugins 2>/dev/null \
  | grep -F "$PLUGIN_NAME" \
  | grep -oE '[0-9]+\.[0-9]+\.[0-9]+[^ ]*' \
  | head -1)

# Latest version on npm
LATEST=$(npm view "$PLUGIN_NAME" version 2>/dev/null)

if [ -z "$LATEST" ]; then
  echo "PLUGIN_NPM_UNAVAILABLE"
elif [ -z "$INSTALLED" ]; then
  csdx plugins:install "$PLUGIN_NAME" && echo "PLUGIN_INSTALLED:$LATEST" || echo "PLUGIN_INSTALL_FAILED"
elif [ "$INSTALLED" = "$LATEST" ]; then
  echo "PLUGIN_UP_TO_DATE:$INSTALLED"
else
  csdx plugins:uninstall "$PLUGIN_NAME" \
    && csdx plugins:install "$PLUGIN_NAME" \
    && echo "PLUGIN_UPDATED:$INSTALLED→$LATEST" \
    || echo "PLUGIN_UPDATE_FAILED:$INSTALLED→$LATEST"
fi

Interpret the last line:

• PLUGIN_UP_TO_DATE:<version> → already on the latest version; proceed to the Eval. (Stay silent — don't report "already up to date" to the user.)

• PLUGIN_INSTALLED:<version> → freshly installed. Briefly tell the user, then proceed.

• PLUGIN_UPDATED:<old>→<new> → uninstalled old version and installed latest. Briefly tell the user the plugin was updated, then proceed.

• PLUGIN_INSTALL_FAILED / PLUGIN_UPDATE_FAILED:<old>→<new> → the install or reinstall errored. Show the user the error output, then display this message and wait:

  The automatic plugin install failed. Please run the following command manually in your terminal, then click Continue when done:

  csdx plugins:install @contentstack/cli-external-migrate
  

  After the user clicks Continue, skip straight to the Eval below to verify the plugin is working before proceeding.

• PLUGIN_NPM_UNAVAILABLE → npm view returned nothing (network issue or package not yet published). Show the user this message and wait:

  Could not reach npm to check the plugin version. Please run the following command manually in your terminal, then click Continue when done:

  csdx plugins:install @contentstack/cli-external-migrate
  

  After the user clicks Continue, proceed to the Eval below.

Eval — Verify Step 2 before proceeding

csdx migrate --help

Pass criteria:

• Output includes migrate:create (the one-shot command used in Step 3)

If migrate:create is present, update the migration checklist to set cf-step2="completed" and cf-step3="in_progress", then proceed to Step 3. If the command errors or migrate:create is missing, re-run the install block above and re-verify.

Step 3 — Content Migration

[PROGRESS] Output the migration progress block: Steps 1–2 → "completed", Step 3 → "in_progress", Step 4 → "pending".

This single command exports the Contentful space, converts the content to a Contentstack bundle, and imports it into a brand-new stack — all in one shot. The master locale is auto-detected from the export's default locale.

3.0 — Retrieve the stored Management Token

The Contentful CLI stores the token from the login in ~/.contentfulrc.json. Read it now so migrate:create can use it without prompting:

$PYTHON_CMD -c "
import json, sys, pathlib
for p in ['~/.contentfulrc.json', '~/.config/contentful/config.json']:
    f = pathlib.Path(p).expanduser()
    if f.exists():
        d = json.loads(f.read_text())
        tok = d.get('managementToken') or d.get('cmaToken') or d.get('management_token')
        if tok:
            print(tok)
            sys.exit(0)
print('NOT_FOUND', file=sys.stderr)
sys.exit(1)
"

Capture the printed value as CONTENTFUL_MANAGEMENT_TOKEN. Do not display it to the user or write it to any file.

If the token is NOT found (exits 1), tell the user the token could not be resolved. Ask them to run contentful login again, then retry.

3.1 — Confirm before running

Tell the user:

"Ready to start the content migration. This will export your Contentful space, convert the content, and import it into a new stack under your org. Shall I proceed?"

Wait for confirmation before running the command.

3.2 — Run migrate:create (output captured to log file)

Run from $SESSION_DIR so that the import logs/ directory and _backup_*/ are written there, keeping the user's project directory clean. Pipe through tee so output streams live and is also saved for parsing:

cd "$SESSION_DIR" && \
csdx migrate:create --source contentful \
  --space-id "$SPACE_ID" \
  --source-token "$CONTENTFUL_MANAGEMENT_TOKEN" \
  --org "$ORG_UID" \
  --download-assets \
  --output "$SESSION_DIR" \
  --workspace "$SESSION_DIR" \
  -y \
  2>&1 | tee "$SESSION_DIR/migrate-create.log"

Flag notes:

• --space-id / --source-token — Contentful source (token resolved in 3.0)
• --org — the org UID captured in Step 1; a new stack is created here
• --source contentful — declares Contentful as the migration source
• --download-assets — include asset binaries in the migration
• --output "$SESSION_DIR" — bundle written to $SESSION_DIR/bundle/
• --workspace "$SESSION_DIR" — export JSON saved to $SESSION_DIR/export.json
• -y — skip internal confirmation prompts (we already confirmed above)
• cd "$SESSION_DIR" — ensures logs/ and _backup_*/ land in the session dir, not CWD

The command runs three phases and prints this progression:

Phase 1 — Export (streams fetch progress, ends with the entity-count table):

┌────────────────────────┐
│ Exported entities      │
├───────────────────┬────┤
│ Content Types     │ 16 │
│ Entries           │ 53 │
│ Assets            │ 21 │
│ Locales           │ 2  │
│ …                 │ …  │
└───────────────────┴────┘
Stored space data to json file at: <SESSION_DIR>/export.json

Phase 2 — Convert (transforms export into Contentstack bundle):

  validate   ✓  export.json
  extract    ✓  2 locales · 16 types
  transform  ✓  53 entries · 13 types  →  <SESSION_DIR>/bundle
  Bundle: <SESSION_DIR>/bundle (16 types, 53 entries)

Phase 3 — Import (creates stack, imports content, ends with the summary box):

✓ Stack created · via cma
──────────────────────────────────────
  Stack name : Contentful Migration 2026-06-08
  Stack key  : blt3e69b8da307655a7
  Region     : AWS-NA
──────────────────────────────────────
… (import progress) …
SUCCESS: Successfully imported the content to the stack named … with the API key blt… .
SUCCESS: The log has been stored at: <SESSION_DIR>/logs
✓ Bundle metadata written: <SESSION_DIR>/bundle/metadata.json
✓ Migration complete
──────────────────────────────────────
  Stack name : Contentful Migration 2026-06-08
  Stack key  : blt3e69b8da307655a7
  Region     : AWS-NA
──────────────────────────────────────

3.3 — Parse the output and extract session variables

Once the command returns, parse the captured log with exact patterns matching the real output:

$PYTHON_CMD - <<'PYEOF'
import re, pathlib, os

raw = pathlib.Path(os.environ["SESSION_DIR"] + "/migrate-create.log").read_text()
# Strip ANSI escape codes — the import phase wraps paths in color sequences
log = re.sub(r'\x1b\[[0-9;]*m', '', raw)

# Patterns matched against real command output
export_json = re.search(r'Stored space data to json file at:\s+(\S+)',  log)
bundle_dir  = re.search(r'Bundle:\s+(\S+)',                             log)
stack_name  = re.findall(r'Stack name\s*:\s*(.+)',                      log)
stack_key   = re.findall(r'Stack key\s*:\s*(blt\w+)',                   log)
region      = re.findall(r'Region\s*:\s*(\S+)',                         log)
log_dir     = re.search(r'SUCCESS: The log has been stored at:\s+(\S+)',log)
metadata    = re.search(r'Bundle metadata written:\s+(\S+)',            log)

session_dir = os.environ["SESSION_DIR"]
# Use last match for stack fields — the final summary box is authoritative
print("EXPORT_JSON="   + (export_json.group(1).strip() if export_json else session_dir + "/export.json"))
print("BUNDLE_DIR="    + (bundle_dir.group(1).strip()  if bundle_dir  else session_dir + "/bundle"))
print("STACK_NAME="    + (stack_name[-1].strip()       if stack_name  else "UNKNOWN"))
print("STACK_KEY="     + (stack_key[-1].strip()        if stack_key   else "UNKNOWN"))
print("REGION="        + (region[-1].strip()           if region      else "UNKNOWN"))
print("LOG_DIR="       + (log_dir.group(1).strip()     if log_dir     else "UNKNOWN"))
print("METADATA_PATH=" + (metadata.group(1).strip()    if metadata    else session_dir + "/bundle/metadata.json"))
PYEOF

Capture each printed KEY=value as a session variable:

Also set the mapper path (always $BUNDLE_DIR/mapper.json):

MAPPER_PATH="$BUNDLE_DIR/mapper.json"

If any value is UNKNOWN, check $SESSION_DIR/migrate-create.log directly for the line and set it manually before continuing.

3.4 — Show the user a progress summary

Show the entity-count table and the final stack summary, both taken from the captured log:

# Entity count table from Phase 1 (export)
grep -A 18 'Exported entities' "$SESSION_DIR/migrate-create.log" | head -20

# Final stack summary box from Phase 3 (last occurrence)
grep -A 4 'Stack name' "$SESSION_DIR/migrate-create.log" | tail -6

Report in plain prose:

"Content migration complete — stack $STACK_NAME ($STACK_KEY, $REGION) is ready."

Handle token expiry mid-run

The Contentstack OAuth token can expire during a long import. The symptom in the log is:

stack creation failed.
  CMA:  401 The provided access token is invalid or expired or revoked

Re-authenticate and re-run from step 3.1:

csdx auth:login --oauth

The command is safe to re-run — it creates a fresh stack each time.

Eval — Verify Step 3 before proceeding

Confirm the key artifacts exist:

ls -lh "$METADATA_PATH"
ls -lh "$MAPPER_PATH"

Run the bundled import summary parser to verify entity counts match what was exported:

$PYTHON_CMD "$SKILL_DIR/scripts/parse_import_summary.py" "$LOG_DIR" --export "$EXPORT_JSON"

Pass criteria:

• STACK_KEY starts with blt (a real stack was created)
• METADATA_PATH and MAPPER_PATH both exist
• SUCCESS: Successfully imported… line is present in the captured log
• Imported counts match exported counts; any divergence must be explained before proceeding

Present the counts table to the user:

Module          Imported   (Exported)
Locales         2          2
Content Types   16         16
Assets          21         21
Entries         53         53
✓ Imported into stack "<STACK_NAME>" (<STACK_KEY>)

If all pass, update the migration checklist to set cf-step3="completed" and cf-step4="in_progress", then proceed to Step 4. If counts diverge or artifacts are missing, surface the discrepancy and point the user at $LOG_DIR — do not proceed until resolved.

Session variables carried into Step 4

Step 4 — Code Migration

[PROGRESS] Output the migration progress block: Steps 1–3 → "completed", Step 4 → "in_progress".

With the content now in Contentstack, migrate the application code that reads from the CMS. This step runs a full detect → plan → rewrite → eval cycle with 13 post-migration checks.

4.1 — Collect inputs

repoPath — local file system path to the application to migrate. Ask the user:

"What is the local path to the codebase you want to migrate? (e.g. /Users/you/projects/my-app)" If it's a remote git repo, clone it first:

cd "$SESSION_DIR" && git clone <REPO_URL>

Then use the cloned directory as repoPath.

mapperPath — the field-mapping JSON produced by Step 3. Confirm it exists:

ls -lh "$SESSION_DIR/bundle/mapper.json"

If the file is at a different path (e.g. the user used a custom workspace), ask them to confirm the exact path.

Read mapperPath with the Read tool and extract the fieldMapping arrays from each content type to build the Contentful-field-UID → Contentstack-field-UID table. Present the extracted table to the user for confirmation before proceeding.

metadataPath — the credentials JSON written by Step 3's import. Confirm it exists and read it:

ls -lh "$SESSION_DIR/bundle/metadata.json"

Read metadataPath with the Read tool. It supplies the new stack's Delivery SDK credentials so you do not have to ask the user for them:

The region (e.g. AWS-NA) comes from csdx config:get:region. Treat the tokens as secrets — never echo them back or write them into files in this workspace; only set them in the migrated app's local .env (which should be gitignored). If metadata.json is missing, fall back to asking the user for CS_API_KEY, CS_DELIVERY_TOKEN, and CS_ENVIRONMENT.

4.2 — Logging

First resolve {SKILL_DIR} (see the "Bundled scripts & references" operating principle) — the directory this skill is installed in — and set it as a shell variable so every command below can reach the bundled scripts portably, on any OS or assistant:

SKILL_DIR="<absolute path to this skill's install dir>"   # the folder holding this SKILL.md

From the first action in this step, record everything to the session log using "$SKILL_DIR/scripts/log.sh":

• log.sh <target> user-input "<what the user asked/answered>" — every user input/communication.
• log.sh <target> decision "<detection results, choices, assumptions, guessed UIDs>".
• log.sh <target> ai-action "<files/queries you changed>" and communication "<what you told the user>".
• log.sh <target> exception "<any error, blocker, or uncertainty you hit>" — log all exceptions.
• Run shell commands through it so output + exit codes are captured and failures auto-log as exceptions: log.sh <target> run "typecheck" -- npx tsc --noEmit.

End with log.sh <target> summary. The log lands in <target>/.migration/ (session.log + session.jsonl + per-command output). See "$SKILL_DIR/scripts/README.md".

4.3 — Reference (read first)

The complete, source-verified mapping for every API, response shape, query operator, rich-text renderer, asset transform, GraphQL query, and Live Preview API lives in:

{SKILL_DIR}/references/CONTENTFUL_TO_CONTENTSTACK_MIGRATION_CONTEXT.md (read it with the Read tool from this skill's install dir — see the {SKILL_DIR} operating principle)

Read it in full before doing anything. It is the single source of truth — follow it; do not rely on prior knowledge where the doc is specific. Key map:

• §0.1 — detect the data-access approach (decision table)
• §1–§16 — REST Delivery SDK (@contentstack/delivery-sdk) mapping
• §17 — GraphQL Content API migration
• §18 — Live Preview / draft mode migration
• §19 — raw REST / fetch and framework source plugins
• §13 gotchas, §15 checklist

4.4 — Procedure

1. DETECT (doc §0.1). Determine the app's language, framework, and which data-access approach(es) it uses, plus whether it implements Live Preview / draft mode. Migrate in the SAME language and framework, preserving the SAME approach (REST→REST, GraphQL→GraphQL, preview→preview). Report findings and PAUSE for confirmation.

2. PREREQUISITES. Confirm the target Contentstack stack already has the matching content model and published content, and confirm the Contentful-field-ID → Contentstack-field-UID map from step 4.1. If any UIDs are unknown, infer from a sample Contentstack entry and FLAG every guess. Confirm env vars (CS_API_KEY, CS_DELIVERY_TOKEN, CS_ENVIRONMENT, + region/branch, + preview token if Live Preview) — these come from metadataPath (step 4.1), not the user (use stack_api_key, NOT stack_id, for CS_API_KEY). If the content model or entries do not yet exist in Contentstack, STOP — code migration requires content to be imported first (Steps 1–3).

3. PLAN. Produce a table: file:line → source call → Contentstack equivalent (cite the doc section) → field-UID dependencies → risk notes. Show it and PAUSE before editing any file.

4. MIGRATE per the doc, following the section matching each detected approach:

   • REST Delivery SDK: §1–§16. GraphQL: §17. Raw REST / framework plugins: §19.
   • Rich text / assets / locales / pagination: §9–§12. Make minimal, mechanical edits that match surrounding code style and the framework's existing data-fetching idioms. Preserve function/component contracts.

5. LIVE PREVIEW. If step 1 found preview/draft-mode, reimplement it per §18, matching the source's scope (routes, components, SSR vs client, click-to-edit vs read-only). Keep preview tokens server-side and preserve existing preview gating/routing.

6. VERIFY — run the eval suite (this is mandatory, not optional). Install deps first so the build eval is meaningful, then run the bundled evals in parallel:

   bash "$SKILL_DIR/scripts/run-all.sh" <path-to-migrated-app>
   

   For maximum parallelism you may instead spawn one agent per "$SKILL_DIR/scripts/"NN_*.sh. See "$SKILL_DIR/scripts/README.md" for what each check catches and exit-code semantics.

   • Hard-gate FAIL/ERROR (residue, field-access, sdk-init, build, secrets) ⇒ the migration is NOT done. Fix and re-run until they pass.
   • Triage every review-eval finding at its file:line — static greps flag suspects, not proven bugs. Fix the true positives; state explicitly why any remaining ones are safe. Never dismiss findings silently.
   • A green build is necessary but not sufficient — it cannot catch reference-array bugs, wrong field UIDs, or RTE output. Still smoke-test live queries against the real Contentstack stack.
   • For anything the doc marks "verify against current docs" (GraphQL hosts/headers, Live Preview front-end API), confirm before finalizing rather than guessing.

4.5 — Guardrails

• Do NOT modify content modeling or move/import entries — out of scope for this step.
• Convert every reference dereference to safe array access (?.[0]) and audit null safety (Contentstack resolves references to arrays, not single objects).
• When a UID or behavior is uncertain, leave a // TODO(migration): comment and list it — never guess silently.

4.6 — Wrap up

[PROGRESS] Output the migration progress block: Steps 1–4 all → "completed". Then immediately proceed to Step 5.

Give the user a final summary of the whole migration journey:

• Content: counts imported into the stack (from Step 3).
• Code: files changed, the rich-text rendering strategy chosen, guessed UIDs to verify, and any TODO(migration) call sites still needing attention.
• Eval results: whether all hard gates passed and which review findings remain to triage.
• **Detection summar

…(truncated)