Import memory entries from a markdown export file, merging with MEMORY.md without duplicating.
The first argument is the project path. Before doing anything else:
Imports memory entries from a markdown file (produced by /skill cmd_instinct_export or written manually) into the workspace MEMORY.md. Duplicate sections are detected and skipped. New sections are appended.
/skill cmd_instinct_import <project-path> <file-path>
/skill cmd_instinct_import <project-path> https://example.com/memory.md
/skill cmd_instinct_import <project-path> team-memory.md --dry-run
/skill cmd_instinct_import <project-path> team-memory.md --force
--dry-run — show what would be added without writing anything--force — skip the confirmation promptRead the import source:
http:// or https://, fetch it:
curl -s "<url>"
Parse import sections: extract all ##-level headings and their content blocks.
Skip export header lines (# Memory Export, # Exported:, # Source:,
# Sections:, and the --- separator).
Read the existing ~/.openclaw/workspace-main/MEMORY.md.
Extract all existing ##-level heading names for deduplication.
Classify each imported section:
Report the analysis:
Import Analysis: <source>
================================
Found <N> sections in import file.
New sections (<count>):
+ <heading name>
+ <heading name>
Duplicate sections (<count>, will be skipped):
= <heading name>
If --dry-run is set, stop here.
If there are no new sections, report "Nothing to import — all sections already exist." Stop.
If --force is not set, ask: "Import <count> new sections? (yes/no)"
If the user declines, stop.
Append the new sections to ~/.openclaw/workspace-main/MEMORY.md.
Re-index memory:
openclaw memory index
Report:
Import complete.
Added: <count> sections
Skipped: <count> duplicates
MEMORY.md now has <total> sections.