Upload files to Memory Brain

How Lead360 helps

Memory Brain file upload turns approved YAML, JSON, Markdown, text, CSV, and text-based PDF content into governed pending revisions. The batch validates together, rejects credentials, preserves file and source evidence, and prevents a partially accepted upload from leaving an unclear knowledge state.

When to use this workflow

Use file upload for stable business knowledge that is not available through a supported Lead360 sync and has an authorized source owner.

Before you begin

Start with the exact sub-account and the smallest set of permissions, records, and resources required for the task.

• Remove passwords, tokens, API keys, payment credentials, and unnecessary personal data.
• Use a supported text-based format and clear file names.
• Confirm source ownership, accuracy, retention, and intended consumer scope.

Complete the workflow

Follow the steps in order and stop when a required resource, permission, review, or approval is missing.

1. Open Memory Brain and choose file or folder upload.
2. Select supported YAML, JSON, Markdown, text, CSV, or text-based PDF files.
3. Review batch validation, detected credentials, normalization, and file count.
4. Submit the valid batch to create pending revisions.
5. Review each normalized revision before making any snapshot available.

Understand the key controls

Atomic batch

The batch is accepted as a unit. A validation failure prevents a partial upload that could hide missing context.

Credential rejection

Detected credentials are rejected because memory is not a secret store.

Text-based PDF

A PDF must contain extractable text. Scanned image-only documents require an approved text preparation process before upload.

Confirm the result

Do not rely on a success message alone. Confirm the saved state and any live effect in the product area that owns the final record.

• Confirm every intended file appears once as a pending revision.
• Compare normalized content with the source.
• Review checksum, evidence, and sensitive-data handling before approval.

Troubleshooting

The batch is rejected

Open the validation details and correct unsupported type, unreadable content, credential detection, duplicate path, size, or normalization errors.

A PDF produces little text

Confirm it contains a text layer. Prepare an accessible text or Markdown source rather than relying on an image-only scan.