Reference for the Skene journey workflow and related commands. Checked against OSS source revision 8eb7942 on 12 October 2026. The published package is 0.5.1; live database input uses the pinned source command below.

The registered commands are analyze, analyse-journey, plan, build, status, push, config, validate, login, logout, and features. Run uvx skene <command> --help for options outside this workflow. The current source does not register serve or attach.

For in-depth usage of individual commands, see the guides. This page is a lookup reference.

Global Options

Flag	Description
`--version`, `-V`	Show version and exit
`--help`	Show help message and exit

When invoked with no arguments, skene prints help and exits.

`analyse-journey`

Generate a journey.yaml describing the user lifecycle of a product.

The journey pipeline runs code and schema agents in parallel when their inputs are supplied. Each proposes candidate milestones. A deterministic merge step with no model in the loop reconciles the two sets, a per-milestone classification pass places them, and the run assembles one validated Customer Journey across seven lifecycle stages: discovery, onboarding, activation, engagement, retention, expansion, and virality. The only artifact written into your project is skene-context/journey.yaml. The CLI executes the journey pipeline locally using the configured model provider. A remote provider receives analysis context.

skene analyse-journey [PATH] [OPTIONS]

Arguments

Argument	Default	Description
`PATH`	none	Path to codebase directory. Pass `.` for the current directory; omit only when supplying a schema input.

Options

Flag	Short	Default	Description
`--schema-dir PATH`			Directory of pre-exported `*.sql` files for the schema agent
`--db-url TEXT`			PostgreSQL connection string for live schema introspection (source revision only; alternative to `--schema-dir`; also `SKENE_DB_URL`)
`--output PATH`	`-o`	`./skene-context/journey.yaml`	Output path for `journey.yaml`
`--product-name TEXT`			Product name in the output (default: inferred from repo directory name, DB name, or schema dir name)
`--api-key TEXT`		`$SKENE_API_KEY` or config	API key for the LLM provider
`--provider TEXT`	`-p`	config value	LLM provider: `openai`, `gemini`, `anthropic`/`claude`, `lmstudio`, `ollama`, `generic`, `skene`
`--model TEXT`	`-m`	provider default	LLM model name
`--base-url TEXT`		`$SKENE_BASE_URL` or config	Base URL for API endpoint
`--schema-max-turns INT`		`150`	Maximum agent turns for the schema agent (range: 1–500)
`--code-max-turns INT`		`200`	Maximum agent turns for the code agent (range: 1–500)
`--classify-concurrency INT`		`8`	Parallel classifier requests (range: 1–64)
`--no-specialize`		`false`	Skip stage specialization; use canonical stage vocabulary
`--quiet`	`-q`	`false`	Suppress status messages; show only errors and final results
`--debug`		`false`	Show diagnostic messages and log all LLM input/output
`--no-fallback`		`false`	Disable model fallback on rate limits; retry same model instead

Schema sources

The schema agent requires one of two inputs, never both:

SQL files (--schema-dir): Provide a directory of pre-exported *.sql files defining tables, views, constraints, and indexes.
Live database (--db-url): Connect directly to a running PostgreSQL database. Skene introspects all user-defined schemas at runtime (excluding pg_catalog, information_schema, pg_toast, and any schema prefixed with pg_). Credentials are never stored.

--schema-dir and --db-url are mutually exclusive. At least one of PATH, --schema-dir, or --db-url must be provided. When PATH is omitted, only the schema agent runs (no code agent).

Behavior notes

Requires a configured LLM (API key + provider). Local providers (lmstudio, ollama, generic) do not require an API key.
The generic provider requires --base-url.
When --db-url is used without --product-name, the database name is extracted from the connection string for the product name.
--db-url credentials are used live and never persisted; anything stored or streamed shows a redacted form.
A normal CLI run writes the local journey. A linked TUI run using the Skene provider can automatically publish the first journey when that workspace has none. See the push guide.

See the CLI help output (skene analyse-journey --help) for detailed usage.

`serve`

This command is not registered in the current OSS source or the published 0.5.1 package. Use analyse-journey for local analysis.

`attach`

This command and remote-server mode are not supported by the current OSS source or published package. See the HTTP API status.

`status`

Show implementation status for skene-context/engine.yaml.

Loads skene-context/engine.yaml, validates structure, and checks whether action-enabled features have matching trigger/function entries in supabase/migrations/*.sql.

skene status [PATH] [OPTIONS]

Arguments

Argument	Default	Description
`PATH`	`.`	Path to the project root directory (must exist)

Options

Flag	Short	Default	Description
`--context PATH`	`-c`		Deprecated. If set to a Skene bundle directory (`skene-context/` or legacy `skene/`), the parent directory is used as project root.
`--find-alternatives`		`false`	Deprecated for engine status checks; currently ignored.
`--api-key TEXT`		`$SKENE_API_KEY` or config	Deprecated for engine status checks; currently ignored.
`--provider TEXT`	`-p`	config value	Deprecated for engine status checks; currently ignored.
`--model TEXT`	`-m`	provider default	Deprecated for engine status checks; currently ignored.

Project root resolution

Uses PATH as project root by default.
If --context points to a Skene bundle directory (.../skene-context or .../skene), uses the parent directory as project root.

Behavior notes

Validates skene-context/engine.yaml and checks duplicate keys/required fields.
For features with action, expects matching trigger/function tokens in SQL migrations. The detail column shows the latest matching migration filename; if the same trigger appears in older files, the suffix (+N) indicates how many additional matches exist.
For features without action, reports code-only mode (no trigger required).
Returns non-zero exit code when required engine/migration checks fail.

See the status guide for detailed usage.

`config`

Manage skene configuration files.

skene config [OPTIONS]

Options

Flag	Default	Description
`--init`	`false`	Create a sample `.skene.config` file in the current directory
`--show`	`false`	Show current configuration values and exit (no interactive editing)

Default behavior (no flags)

When invoked without --init or --show:

Displays current configuration values (same as --show)
Asks whether you want to edit the configuration
If yes, launches an interactive setup flow to select provider, model, and enter an API key

Configuration load order

Configuration is resolved in this order (later sources override earlier ones):

User config: ~/.config/skene/config
Project config: ./.skene.config
Environment variables: SKENE_API_KEY, SKENE_PROVIDER
CLI flags

See the configuration guide for file format and all supported options.

`push`

Upload the Skene bundle (all files under {output_dir}) and the latest trigger migration to Skene Cloud.

push uploads existing files from the bundle. It does not generate them. Engine metadata and the latest trigger migration are included when present; a journey-only bundle does not require them.

skene push [PATH] [OPTIONS]

Arguments

Argument	Default	Description
`PATH`	`.`	Project root. Artifact paths and sticky `output_dir` resolution use this directory (see configuration).

Options

Flag	Short	Default	Description
`--upstream TEXT`	`-u`	config / default cloud	Workspace URL (e.g. `https://skene.ai/workspace/my-app`). Also read from `.skene.config`.
`--quiet`	`-q`	`false`	Suppress non-error output.
`--debug`		`false`	Diagnostic messages and LLM debug logging.

Behavior notes

Includes all existing bundle files and the latest eligible trigger migration when present. Missing engine metadata does not block the upload.
Sends a JSON payload with manifest and files (project-relative paths and file contents) to the upstream /api/v1/push API.
Use skene login (or SKENE_UPSTREAM_API_KEY) for authentication.

See the push guide for detailed usage.

`login`

skene login [OPTIONS]

Options

Flag	Short	Default	Description
`--upstream TEXT`	`-u`		Upstream workspace URL (e.g. `https://skene.ai/workspace/my-app`)
`--status`	`-s`	`false`	Show current login status for this project

Behavior notes

Saves upstream URL, workspace, and API key to .skene.config with restrictive permissions (0600).
Use --status to check whether you are logged in for the current project.

See the login guide for detailed usage.

`logout`

Log out from upstream (remove saved token).

skene logout

Behavior notes

Removes upstream credentials from .skene.config.
Does not invalidate the token server-side.

`features`

Manage the growth feature registry.

`features export`

Export the feature registry for use in external tools.

skene features export [PATH] [OPTIONS]

Arguments

Argument	Default	Description
`PATH`	`.`	Project root (to locate the Skene bundle directory)

Options

Flag	Short	Default	Description
`--context PATH`	`-c`	auto-detected	Path to the Skene bundle directory (`skene-context/` or legacy `skene/`)
`--format TEXT`	`-f`	`json`	Output format: `json`, `csv`, `markdown`
`--output PATH`	`-o`	stdout	Output file path. Prints to stdout if omitted.

Behavior notes

Reads feature-registry.json from the context directory.
The registry is populated by analysis runs; it must exist before exporting.
Use for integrating with dashboards, Linear, Notion, or documentation.

See the features guide for detailed usage.

Environment Variables

Variable	Used by	Description
`SKENE_API_KEY`	`analyse-journey`	API key for the LLM provider. Equivalent to `--api-key`.
`SKENE_BASE_URL`	`analyse-journey`	Base URL for OpenAI-compatible endpoints. Equivalent to `--base-url`.
`SKENE_PROVIDER`	config loading	LLM provider override at the environment level.
`SKENE_OUTPUT_DIR`	all commands	Override `output_dir` for commands that have no dedicated flag (primarily `push`).
`SKENE_UPSTREAM_API_KEY`	`push`, `login`	API key for upstream authentication.
`SKENE_DEBUG`	all commands	Enable debug mode (`true`/`false`).
`SKENE_DB_URL`	`analyse-journey`	PostgreSQL connection string for live schema introspection. Equivalent to `--db-url`.

Exit Codes

Code	Meaning
`0`	Success
`1`	Command failure, such as missing credentials or failed validation
`2`	Invalid command usage or conflicting/missing journey inputs

Examples

# Full workflow
uvx skene config --init
uvx skene config
uvx skene analyse-journey .

# Analyse user journey with SQL schema files
uvx skene analyse-journey ./my-app --schema-dir ./schemas

# Analyse with a live database using the verified source revision
uvx --from git+https://github.com/SkeneTechnologies/skene.git@8eb7942f569a4f99387e5c05b086c7babc8ce3d3 skene analyse-journey --db-url "postgresql://user:pass@localhost:5432/mydb"

# Journey analysis with custom output
uvx skene analyse-journey ./my-app --schema-dir ./schemas -o ./output/journey.yaml

# Journey analysis with explicit provider settings
uvx skene analyse-journey . -p gemini -m gemini-3-flash-preview --api-key "YOUR_KEY"

# Journey analysis with a local LLM (no API key needed)
uvx skene analyse-journey . -p ollama -m llama3

# Check engine implementation status
uvx skene status

# Push artifacts upstream
uvx skene push
uvx skene push --upstream https://skene.ai/workspace/my-app

# Login/logout from upstream
uvx skene login --upstream https://skene.ai/workspace/my-app
uvx skene login --status
uvx skene logout

# Export feature registry
uvx skene features export --format markdown -o features.md

CLI Reference

Global Options

analyse-journey

Arguments

Options

Schema sources

Behavior notes

serve

attach

status

Arguments

Options

Project root resolution

Behavior notes

config

Options

Default behavior (no flags)

Configuration load order

push

Arguments

Options

Behavior notes

login

Options

Behavior notes

logout

Behavior notes

features

features export

Arguments

Options

Behavior notes

Environment Variables

Exit Codes

Examples

`analyse-journey`

`serve`

`attach`

`status`

`config`

`push`

`login`

`logout`

`features`

`features export`