Skene OSS
MIT licencePython 3.11+Skene OSS is for developers reading their codebase, not the growth and revenue teams who buy Skene Cloud. The MIT-licensed CLI and terminal UI at github.com/SkeneTechnologies/skene. Free, no Skene account.
skene to open it.PostgreSQL connector included
--db-urlintrospects a live database at runtime. Credentials are used live, never stored.Still published and still maintained
MIT licence, Python 3.11+. pip and uvx pull it from PyPI; the TUI binary comes from GitHub Releases.
The MIT licence is what you keep
If Skene Cloud stopped tomorrow the tool still runs, and the journey file is already in your repository.
Read only, unless you hand it a database.
Four things the tool can touch, and the access it has to each.
Your codebase
Read only.
Database schema
Read only, and only if you give it one. One of the two, never both.
Credentials
Used live, never persisted. Redacted anywhere they are displayed.
Its own output
A plain YAML file written into your repository.
Two agents, one deterministic merge.
analyse-journey runs two subagents in parallel: one reads the codebase, one reads the database schema. A deterministic merge with no model in the loop reconciles the two into one customer journey across seven lifecycle stages.
- discovery, onboarding, activation, engagement, retention, expansion, virality
- One artifact:
skene-context/journey.yaml, written into your repository
| Part | What it does |
|---|---|
| Code agent | Walks the repository |
| Schema agent | Walks the database schema |
| Merge | Deterministic, no model in the loop |
| Output | skene-context/journey.yaml |
The PostgreSQL connector.
Two ways to hand it a schema, one of the two, never both. Pass neither and no database is read at all.
--schema-dirtakes exported.sqlfiles--db-urltakes a live PostgreSQL string, introspected at runtime and never persisted- System schemas are excluded; row data is never read
Seven values for --provider.
It selects who handles the analysis model calls. --api-key sets the key. Two of the seven run locally, one routes through Skene.
- openai, gemini, anthropic: with your own API key
- lmstudio, ollama: local, no key, nothing leaves the machine
| --provider | Runs against |
|---|---|
| generic | Any OpenAI-compatible endpoint, --base-url |
| skene | needs Skene account |
It writes one file.
Skene OSS reads a repository, writes a file, and keeps a copy of nothing. Watching releases, reviewing pull requests and keeping a record between runs are Skene Cloud, a different product with an account behind it.
--provider skenepoints the CLI at Skene as a model provider without making the run a Cloud product- The TUI wizard needs no prerequisites and runs the same engine
Run it on your own repository.
The install commands are at the top of this page, and the run needs no Skene account. Reviewing every pull request is Skene Cloud.