Klik Memory Import

Prerequisites

First-time setup: If ~/.claude/skills/klik-import/dist/klik-import.mjs does not exist, run:

git clone https://github.com/minervacap2022/klik-import-skill ~/.claude/skills/klik-import

If already installed, ensure it is up to date:

git -C ~/.claude/skills/klik-import pull --ff-only

Node >= 18 is required (check: node --version).

Run node ~/.claude/skills/klik-import/dist/klik-import.mjs doctor to verify the environment.

Agent Instructions

You are the collector. Your job is to discover, clean, and structure the user's memory data before handing it to the CLI tool for upload.

Step 1: Get import code

Ask the user:

"Please open Klik App > Settings > Import from Agent > Generate Code, then share the 6-digit code with me."

Wait for the code before proceeding.

Step 2: Discover data sources

Use your native file tools (Glob, Read, Bash) to find:

Memory files:

• Start from hints: ~/.claude/projects/*/memory/, ~/.openclaw/projects/*/memory/
• Recursively find MEMORY.md (index file) and any *.md files alongside it
• If the user mentions other agent directories, look there too
• Do NOT read: .ssh/, .aws/, *credentials*, *.env, *token*, *secret*

Scheduled tasks:

• Look for: ~/.claude/scheduled_tasks.json, ~/.openclaw/scheduled_tasks.json
• Also check crontab if user consents: crontab -l 2>/dev/null

Step 3: Clean before packaging

For each memory file:

• Parse YAML frontmatter (between --- delimiters) into frontmatter field
• Put body text into content field
• Skip files with no meaningful content (empty, only whitespace, only metadata)
• Skip cache files, session transcripts, binary blobs

For each scheduled task:

• Keep: cron, prompt, durable, recurring, created_at
• Skip tasks with no prompt or empty prompt
• Skip tasks where durable is false (they are ephemeral and likely expired)

Step 4: Build draft payload JSON

Construct the payload strictly matching this schema:

{
  "schema_version": "1.0",
  "client": {
    "skill_version": "0.1.0",
    "host_agent": "<your agent name, e.g. claude-code or openclaw>",
    "host_agent_version": "<your version if known, else unknown>",
    "os": "<linux|darwin|win32>",
    "collected_at": "<ISO 8601 UTC timestamp>"
  },
  "redaction": {
    "enabled": true,
    "rules_version": "1.0",
    "redacted_count": 0
  },
  "collectors": [
    {
      "name": "claude_memory",
      "source_root": "<absolute root path>",
      "items": [
        {
          "relative_path": "<path relative to source_root>",
          "type": "markdown_index | markdown_memory",
          "size_bytes": 412,
          "mtime": "<ISO 8601>",
          "frontmatter": { "name": "...", "description": "...", "type": "..." },
          "content": "<cleaned body text>"
        }
      ]
    },
    {
      "name": "scheduled_tasks",
      "source_root": "<directory containing scheduled_tasks.json>",
      "items": [
        {
          "relative_path": "scheduled_tasks.json",
          "type": "scheduled_task",
          "cron": "0 9 * * 1-5",
          "prompt": "...",
          "durable": true,
          "recurring": true,
          "size_bytes": 80,
          "mtime": "<ISO 8601>"
        }
      ]
    }
  ]
}

Write this JSON to a temporary file: /tmp/klik_import_draft.json

Validate before upload:

node ~/.claude/skills/klik-import/dist/klik-import.mjs validate \
     --input /tmp/klik_import_draft.json

If validation fails, fix the issues reported and re-validate.

Step 5: Show summary to user

Before uploading, show the user:

• How many files were found per collector
• Total size in KB
• Any files that were skipped and why

Ask for confirmation to proceed.

Step 6: Upload

node ~/.claude/skills/klik-import/dist/klik-import.mjs submit \
     --input /tmp/klik_import_draft.json \
     --code <6-digit-code>

Step 7: Cleanup

rm -f /tmp/klik_import_draft.json

Tell the user the import ID and that Klik will process the data shortly.