Memory commands

memory

Check, list, locate, and repair the shared memory in the working directory. Each subcommand reads project memory in .agnostic-ai/memory/, then personal memory in .agnostic-ai/local/memory/, or in the personal folder shared across worktrees with memory.personal: repo. Missing memory folders are skipped.

agnostic-ai memory lint     # run only the memory checks
agnostic-ai memory index    # rebuild each MEMORY.md from its fact files
agnostic-ai memory list     # print each fact's scope, type, and title
agnostic-ai memory path     # print each memory folder's absolute path
agnostic-ai memory import --from claude   # copy facts from another memory store

• memory lint reports LINT039 to LINT044, the memory findings lint and doctor also report. It exits 1 on an error, or on a warning with --strict. --json prints the same format as lint --json, with command set to memory lint.
• memory index drops merge conflict markers, links to missing files, and repeated links. Every other line stays in place, headings and notes included. It then adds a - [name](file.md): description line from the YAML fields at the top of each fact without one. Run it after two tools edit the index at once.
• The type comes from metadata.type, or from a top-level type as Claude Code's auto memory writes it.
• With no memory folder, memory lint says so and exits 0.
• memory list prints facts in index order. The title is the index line's link text, or the fact's name when no line links it.
• memory path prints one scope folder line per memory folder, project first, as absolute paths. It works from any folder inside the project and from a linked worktree, and prints a folder that does not exist yet. With memory.personal: repo, every worktree of the repository gets the same personal folder. The shared-memory-policy rule has tools without a session-start hook run it.
• memory import copies fact files from another memory store into personal memory and adds their index lines. --from claude reads Claude Code's own store, the folder it used before sync pointed it at personal memory. --from checkout and --from repo read the other personal folder after a memory.personal switch. --path reads any folder with a MEMORY.md and one file per fact. Every tool reads personal memory, so run --dry-run first: it lists each fact with its description, and --only <file> copies just the ones you name. The source never changes. Lines of the source index that link no fact come along as one more fact (claude-index-notes.md for --from claude). A fact whose file name personal memory already has stays as it is, and a symlink or a file with a line that looks like a secret is skipped; the report names each. A fact already there byte for byte gets its index line back if the index lacks it, so a rerun finishes an import that stopped. Import refuses a personal folder with files Git tracks, and first adds the checkout folder to Git's exclude list when Git would track it. It lists each fact it handled in .imported inside the personal folder, so LINT043 stops counting a fact you delete or merge afterwards. --dry-run prints the plan and writes nothing.