All operations go through the cortex command. Run cortex --help for the
full list.
Create a Python venv and install dependencies.
cortex bootstrap [REPO_ROOT]
| Argument | Default | Description |
|---|---|---|
REPO_ROOT |
auto-detect | Path to the Cortex repo root |
What it does: Creates .venv/ (if missing), upgrades pip, installs
requirements.txt (or the package in editable mode). Idempotent — safe to
re-run.
cortex bootstrap
Set up or upgrade a Cortex installation.
cortex install [VAULT] [--upgrade] [--no-encode] [--dry-run]
| Argument / Flag | Default | Description |
|---|---|---|
VAULT |
prompt | Vault path (interactive prompt if omitted) |
--upgrade |
off | Backup existing files, deploy updates, re-encode |
--no-encode |
off | Skip re-encoding (with --upgrade) |
--dry-run |
off | Preview changes without writing |
What it does (initial install):
- Ensures venv deps are installed (creates venv if needed)
- Deploys encoder scripts to
<vault>/_sync/ - Generates
<vault>/_sync/cortex.yaml(if missing) - Installs the
cortex-aiskill to~/.config/opencode/skills/ - Runs first encoding
cortex install # interactive
cortex install ~/cortex-ai # non-interactive
cortex install --upgrade ~/cortex-ai # upgrade existing
cortex install --dry-run ~/cortex-ai # preview only
cortex install --upgrade --no-encode ~/cortex-ai # skip re-encode
Revert Cortex-installed assets (notes are kept).
cortex uninstall --vault VAULT [--latest] [--backup NAME] [--apply] [--purge]
| Flag | Default | Description |
|---|---|---|
--vault |
prompt | Vault root (required) |
--latest |
true |
Undo only the most recent manifest |
--backup NAME |
None |
Undo a specific backup directory |
--apply |
off | Actually make changes (default is dry-run) |
--purge |
off | Also delete _sync/encoded/ |
cortex uninstall --vault ~/cortex-ai --latest # preview
cortex uninstall --vault ~/cortex-ai --latest --apply # revert
cortex uninstall --vault ~/cortex-ai --latest --apply --purge # full cleanup
Run vault-to-agent encoding.
cortex encode [--dry-run] [--list] [--show-config] [--check] [--graph]
[--purge] [--purge-apply] [--config PATH]
[--hive-push] [--hive-pull] [--hive-status]
| Flag | Description |
|---|---|
--dry-run |
Preview all writes without touching disk |
--list |
List every note with its tier and type |
--show-config |
Print resolved paths as JSON |
--check |
Report release/schema version health |
--graph |
Output wiki-link graph and exit |
--purge |
Preview deletion of drained log/session notes |
--purge-apply |
Delete drained notes and rebuild |
--config PATH |
Path to cortex.yaml (default: cwd/_sync/cortex.yaml) |
--hive-push |
Push vault notes to Cortex Hub |
--hive-pull |
Pull vault notes from Cortex Hub |
--hive-status |
Show hive connection status |
cortex encode # normal run
cortex encode --dry-run # preview
cortex encode --list # see all notes
cortex encode --check # version/schema health
cortex encode --show-config # resolved paths
cortex encode --config ~/cortex-ai/_sync/cortex.yaml # explicit config
Show installation health.
cortex status [--vault VAULT]
| Flag | Default | Description |
|---|---|---|
--vault |
auto-detect | Vault path |
Reports: config found, vault found, encoded memory found, schema version status, opencode config found.
cortex status
cortex status --vault ~/cortex-ai
Validate installation across all platforms.
cortex doctor [--platform NAME] [--vault VAULT]
| Flag | Default | Description |
|---|---|---|
--platform NAME |
all | Check a specific platform only |
--vault |
auto-detect | Vault path |
Checks core assets (config, vault, encoded memory, skill) and each platform integration (opencode, codex, copilot).
cortex doctor
cortex doctor --platform opencode
Import existing agent context into the vault.
cortex import [--vault VAULT] [--dry-run]
[--agents-md PATH] [--claude-md PATH]
[--opencode PATH] [--claude-memory PATH]
| Flag | Description |
|---|---|
--vault |
Target vault (auto-detect if omitted) |
--dry-run |
Preview without writing |
--agents-md |
Path to agents.md |
--claude-md |
Path to CLAUDE.md |
--opencode |
Path to opencode.jsonc |
--claude-memory |
Path to ~/.claude/memory/ directory |
Reads common agent config files, backs them up to
<vault>/_sync/backups/, and seeds the vault with notes tagged review
for manual refinement.
cortex import --dry-run # preview
cortex import # import into auto-detected vault
cortex import --vault ~/cortex-ai # explicit vault
Print version information.
cortex version
Outputs the release version (VERSION file) and schema version
(SCHEMA_VERSION file).
Search encoded memory from the CLI.
cortex memory search QUERY [--vault VAULT]
| Argument / Flag | Description |
|---|---|
QUERY |
Search string (required) |
--vault |
Vault path (auto-detect if omitted) |
Searches memory.json by keyword, ranking across note id, aliases, tags,
category, and body content.
cortex memory search jira
cortex memory search "sprint calendar" --vault ~/cortex-ai
Fetch a single memory note by its id.
cortex memory get ID [--vault VAULT]
| Argument / Flag | Description |
|---|---|
ID |
Note id slug (required) |
--vault |
Vault path (auto-detect if omitted) |
Looks up the note in memory.json first. If not found, falls back to scanning
vault files for <id>.md. Returns full content and metadata.
cortex memory get askdel
cortex memory get jira-rest-api --vault ~/cortex-ai
Lint a Cortex vault for common issues.
cortex lint [--vault VAULT] [--strict] [--fix] [--note ID] [--json] [--rules]
| Flag | Description |
|---|---|
--vault |
Vault path (auto-detect if omitted) |
--strict |
Treat warnings as errors |
--fix |
Auto-fix fixable issues (missing-type, missing-tier, missing-aliases) |
--note ID |
Lint a single note by id |
--json |
Output results as JSON |
--rules |
List available rules and exit |
Rules:
| Severity | Rule | Description | Auto-fix |
|---|---|---|---|
| E | missing-id |
Note is missing an id field in frontmatter |
|
| E | missing-type |
Note is missing a type field in frontmatter |
✓ |
| E | missing-tier |
Note is missing a tier field in frontmatter |
✓ |
| E | invalid-tier |
Note has an unrecognized tier value | |
| E | duplicate-id |
Multiple files share the same id |
|
| W | missing-aliases |
Note is missing an aliases field |
✓ |
| W | slug-mismatch |
Filename stem does not match the id field |
|
| W | dangling-wiki-link |
Wiki-link targets a non-existent note | |
| W | non-slug-id |
id contains non-slug characters |
|
| I | missing-updated |
Note is missing an updated date |
|
| I | empty-body |
Note has no body content |
Exit codes: 0 = clean, 1 = errors found.
cortex lint # scan auto-detected vault
cortex lint --vault ~/cortex-ai # explicit vault
cortex lint --strict # warnings become errors
cortex lint --fix # auto-fix type/tier/aliases
cortex lint --note my-note # single note
cortex lint --json # machine-readable output
cortex lint --rules # list rules
Write a memory note to the vault. Without --body or --body-file, writes
frontmatter only.
cortex memory write --title TITLE --type TYPE --tier TIER
[--tags TAGS] [--category CAT]
[--body TEXT | --body-file PATH]
[--update] [--no-encode] [--vault VAULT]
| Flag | Description |
|---|---|
--title |
Note title (becomes alias and id) |
--type |
Note type: knowledge, entity, feedback, session, log |
--tier |
Tier: core, skill:<name>, project, vault-only |
--tags |
Comma-separated tags |
--category |
Category label |
--body |
Note body content (inline text) |
--body-file |
Path to file containing note body content |
--update |
Patch an existing note (update body + bump date) |
--no-encode |
Skip automatic encode after write (for batch operations) |
--vault |
Vault path (auto-detect if omitted) |
Creates or updates a note file with YAML frontmatter and optional body. By
default triggers cortex encode in the background after writing.
cortex memory write --title "TypeScript Style" --type feedback --tier core
cortex memory write --title "Jira Tips" --type knowledge --tier skill:jira --body "Use JQL for complex queries"
cortex memory write --title "Sprint Notes" --type session --tier vault-only --body-file ./notes.md
cortex memory write --title "Jira Tips" --type knowledge --tier skill:jira --body "Updated content" --update
cortex memory write --title "Batch Item" --type knowledge --tier project --no-encode
Manage agent platform integrations. Each platform supports
install, uninstall, and status.
cortex opencode install [--vault VAULT] [--dry-run]
cortex opencode uninstall [--vault VAULT] [--dry-run]
cortex opencode status [--vault VAULT]
Stub — not yet implemented.
cortex codex install
cortex codex uninstall
cortex codex status
Stub — not yet implemented.
cortex copilot install
cortex copilot uninstall
cortex copilot status