Commands

Authentication precedence (when multiple sources are set): --api-key flag > PLEXT_API_KEY env var > stored session.

All data commands accept --json for machine-readable output.

Setup

`plext setup`

One-shot bootstrap: login → link → start.

plext setup                                                  # interactive
plext setup --api-key plxt_… --plexes a,b --agents claude-code   # non-interactive

OAuth device flow. With --api-key, validates the key and stashes it instead of running OAuth.

plext login
plext login --api-key plxt_…

`plext link`

Link the current project to a workspace; pick which plexes to capture into; pick which agents to hook up.

Replaces the older plext init.

plext link
plext link --plexes decisions,incidents --agents claude-code
plext link --plexes a,b --access a=full,b=read
plext link --workspace my-team --plexes a,b

`plext start`

Install the plext-write and plext-read skills into each configured agent. Run after plext link.

plext start

`plext stop`

Remove the agent skill files. .plext/config.json stays — plext start re-installs.

plext stop

`plext unlink`

Remove .plext/ and all agent skill files. Replaces the older plext cleanup.

plext unlink

`plext status`

Show the current setup — workspace, plexes (with access levels), agents.

plext status

Auth

`plext logout`

Revoke session and delete local credentials.

plext logout

`plext whoami`

Show the logged-in user and the auth source (session vs API key).

plext whoami

Plexes

`plext plex list`

List plexes linked to this project.

plext plex list
plext plex list --json

`plext plex show <slug>`

Show a linked plex's details — config, member count, note count, your role.

plext plex show decisions
plext plex show decisions --json

`plext plex search <query>`

Search linked plexes by name or description.

plext plex search incident

`plext plex create <name>`

Create a new plex and add it to .plext/config.json.

plext plex create architecture-decisions \
  --description "ADRs for the platform team" \
  --visibility public

plext plex create incidents \
  --schema '{"required_fields":[{"key":"severity","type":"select","options":["low","high"]}]}'

Flags: --description, --schema (JSON), --private, --visibility public|private.

`plext plex update <slug>`

Update name, description, schema, or visibility. Same flags as plex create.

plext plex update incidents --description "Production incidents only"

Plex deletion is web-only — destroying a plex destroys its notes, which is too dangerous for a one-line CLI command. Delete from the dashboard instead.

Notes

`plext note create --plex <slug>`

Save a note. This is what the agent skill calls under the hood.

plext note create --plex decisions \
  --title "Chose Postgres over DynamoDB" \
  --body "Need transactions; access patterns aren't truly key-value." \
  --tags database,architecture \
  --field status=accepted

# Body from stdin
echo "..." | plext note create --plex decisions --title "..." --body-file -

# Body from file
plext note create --plex decisions --title "..." --body-file ./body.md

Flags: --title, --body, --body-file, --field key=value (repeatable), --tags, --json.

`plext note show <id> --plex <slug>`

Show full body and metadata.

plext note show note_abc --plex decisions
plext note show note_abc --plex decisions --json

`plext note update <id> --plex <slug>`

Update title, body, tags, metadata.

plext note update note_abc --plex decisions --tags database,architecture,verified

`plext note delete <id> --plex <slug>`

Delete a note. Permanent.

plext note delete note_abc --plex decisions

`plext note search <q>`

Search notes. Across every linked plex by default, or scope with --plex.

plext note search "session tokens"
plext note search "session tokens" --plex security --limit 5

`plext note list --plex <slug>`

List notes in a plex.

plext note list --plex decisions --limit 20
plext note list --plex decisions --json

Global flags

Flag	Notes
`--workspace <slug>`	Override the project's linked workspace for this invocation
`--api-key plxt_…`	Per-invocation auth override
`--json`	Machine-readable output (every data command)
`--help`	Subcommand help

Environment variables

Var	Effect
`PLEXT_API_KEY`	Default API key (overridden by `--api-key`)

Local-only commands work when stopped

plex list, plex show, plex search are local reads — they work even when the project is stopped (plext stop). Data commands (note create, etc.) require plext start.

On this page