feat: simplified flow, push upstream, output-dir consistency by teppohudsson · Pull Request #72 · SkeneTechnologies/skene

teppohudsson · 2026-04-28T04:58:58Z

Summary

Two main themes on this branch:

1. User journey analysis (`user-journey.yaml` / TTV)

schema_analysis.ttv_journey_by_subject is now produced from schema + growth manifest only — you do not need engine.yaml to generate the journey graph used by the visualiser.

Input	Role
`schema.yaml`	Authoritative tables, columns, relationships
`growth-manifest.json`	`current_growth_features`, `growth_opportunities`, and project metadata

Flow: one LLM call returns a single global Time-to-Value DAG (subjects, nodes, edges, valueProxies, lifecycleDataExplanation). The compiler validates it (non-empty subjects/nodes/edges, DAG acyclicity, dataChange on milestone edges, valueProxies linked to real nodes, default stateScope when missing), then groups by subject into the existing storage shape (ordered_nodes, value_nodes, ordered_edges) so the TUI keeps working unchanged.

Optional: if engine.yaml contains planned features, they are still compiled into compiled_features in a separate step.

CLI: skene analyse-user-journey — defaults follow ./skene-context/ (or configured output_dir).

2. Push: entire Skene bundle directory (e.g. `skene-context/`)

skene push uploads all files under the resolved bundle directory (configured output_dir / SKENE_OUTPUT_DIR, default ./skene-context/), not a hard-coded short list. The implementation walks the directory recursively and includes every file as { path, content } in the push payload, plus the latest Skene trigger migration under supabase/migrations/ (and the engine file if it lives outside the bundle).

Why: so Skene Cloud receives the full analysis bundle — schema, growth manifest, user-journey.yaml, feature registry, plans, and any other artifacts written under that folder — in one deploy.

Test plan

uv run pytest tests/ -x -q
uv run ruff check src/ tests/
cd tui && go test ./... && go build ./cmd/skene/

Checklist

Tests added or updated for behavioral changes where needed
Documentation / CLI help updated where behavior changed
Linting and tests pass locally

…dates Push posts files[] with full skene bundle plus trigger migration; align API base and success output with upstream artifact merge. Add journey compiler and analyse user journey; extend plan/journey pipelines. TUI: outputdirs, growth engine and views, visualizer HTML. Docs and tests for upstream. Made-with: Cursor

Unify output_dir: skene-context as default bundle, resolveOutputDir in TUI. Push uploads full bundle; prefer explicit output_dir; preflight engine, trigger, registry warnings. Fix output_paths to match consistency tests. Made-with: Cursor

Made-with: Cursor

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 4325c53d26

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

chatgpt-codex-connector · 2026-04-28T05:02:47Z

+        resolve_path=True,
+    ),
+    schema: Path = typer.Option(
+        Path("./skene/schema.yaml"),


Use skene-context defaults for analyse-user-journey inputs

The new analyse-user-journey command still defaults to ./skene/..., while this commit changes the pipeline defaults to ./skene-context/...; as a result, running skene analyse-user-journey with no flags on a freshly generated project will fail to find schema.yaml/growth-manifest.json/engine.yaml unless users manually override paths. This breaks the zero-config default flow introduced elsewhere in the same change.

Useful? React with 👍 / 👎.

chatgpt-codex-connector · 2026-04-28T05:02:47Z

+        existing = resolve_bundle_dir(Path.cwd())
+        if existing is not None:
+            return f"./{existing.name}"


Resolve sticky output_dir against the target project root

The sticky legacy detection now uses Path.cwd() to pick ./skene vs ./skene-context, which is not the same as the project path passed to commands like skene push /some/project; if the caller’s current directory has a different bundle layout than the target project, config.output_dir can point to the wrong directory and downstream artifact resolution uses the wrong engine path.

Useful? React with 👍 / 👎.

chatgpt-codex-connector · 2026-04-28T05:02:47Z

+    "\n\nIMPORTANT: Your previous response could not be parsed as JSON. "
+    "Return ONLY a single JSON object, no prose, no markdown fences, no trailing commas. "
+    "It must contain exactly the keys ``compiled_features`` and ``ttv_journey_by_subject``."
+)


Align retry reminder with the actual prompt schema

After JSON parse failures, retries append a reminder demanding an object with exactly compiled_features and ttv_journey_by_subject, but both retry call sites expect single feature/subject objects instead; this contradictory instruction can make second-attempt outputs structurally wrong, leading to silently dropped or default-filled compiled entries rather than recovering valid data.

Useful? React with 👍 / 👎.

Made-with: Cursor

…I lint - Resolve bundle layout for Config.output_dir against CLI project_root via bundle_resolution_root and resolve_cli_config(project_root=...). - Align user-journey LLM retry reminder with per-feature/subject JSON shape. - Remove unused getOutputDir wrapper in TUI (golangci-lint unused). Made-with: Cursor

Update configuration, index, llms.txt, python API, and troubleshooting to stay consistent with the simplified flow. Made-with: Cursor

…-journey - Pipeline: journey stage only requires schema + growth manifest - analyse-journey: skip plan but still run journey when growth runs - analyse-user-journey: allow missing engine.yaml (empty engine) - journey_compiler: remove empty-engine warning for TUI - growth_from_schema: reduce MAX_HYPOTHESES 12 → 5 Made-with: Cursor

Single LLM call outputs the TTV DAG (subjects, nodes, edges, valueProxies); validate, then group into ttv_journey_by_subject. Engine features are optional and only fill compiled_features. Update analyse-user-journey help. Made-with: Cursor

Made-with: Cursor

teppohudsson · 2026-04-28T09:06:38Z

technically this works. test pass. I get to run and the CLI push command works. ready to merge

When output_dir is set in config or passed to collect_push_files, use that path verbatim instead of falling back to resolve_bundle_dir when the directory is missing. Sticky bundle discovery applies only when output_dir is unset. Made-with: Cursor

skene push no longer fails when engine.yaml is missing; upload whatever bundle files exist. Engine metadata is only read when the file is present. Removes the hard requirement for a trigger migration so journey-only workflows can deploy without running build first. Made-with: Cursor

bmichele

After testing the analysis with several settings and projects and viewing the journey, pressing n for next steps shows "Deploy to Skene Cloud" greyed out and "not available". This blocks reaching skene push from the TUI entirely.

Steps to reproduce:

cd tui && make clean run-local
Analyse a project
View Journey → press n
"Deploy to Skene Cloud" is disabled

… requiring engine.yaml

bmichele · 2026-04-30T15:32:51Z

I have fixed the issue reported in #72 (review)

I end-to-end tested the user journey generation, i.e.

analyze local project with TUI -> OK
visualize user journey in browser from local files -> OK
push to skene cloud from TUI -> OK
Connect to cloud and "Compile skene engine" from workspace -> DOES NOTHING

The problem is now on the cloud side, so this is good to be merged for me.

bmichele

There is still an issue when compiling the engine on the cloud, but the CLI/TUI part is now fixed: all files that we expect are generated from the analysis and gets pushed to the cloud.

Good to be merged for me

teppohudsson added 3 commits April 27, 2026 23:29

fix(cli): quiet push output — drop preflight spam, keep artifact summary

4325c53

Made-with: Cursor

teppohudsson requested review from bmichele and eretjohannes as code owners April 28, 2026 04:58

github-code-quality Bot found potential problems Apr 28, 2026

View reviewed changes

Comment thread src/skene/cli/commands/push.py Fixed

chatgpt-codex-connector Bot reviewed Apr 28, 2026

View reviewed changes

teppohudsson added 6 commits April 28, 2026 08:08

style: ruff format and fix lint (E501, import order)

abbaf6a

Made-with: Cursor

docs: refresh push guide, CLI reference, and related pages

23aeec7

Update configuration, index, llms.txt, python API, and troubleshooting to stay consistent with the simplified flow. Made-with: Cursor

style(journey): ruff format and shorten TTV prompt line (E501)

9db34df

Made-with: Cursor

teppohudsson and others added 3 commits April 28, 2026 12:28

Merge branch 'main' into feat/simplified-flow

d1a3caa

bmichele requested changes Apr 30, 2026

View reviewed changes

"deploy to skene cloud" in TUI push content of skene-context, without…

d219817

… requiring engine.yaml

bmichele self-requested a review April 30, 2026 15:33

bmichele approved these changes Apr 30, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat: simplified flow, push upstream, output-dir consistency#72

feat: simplified flow, push upstream, output-dir consistency#72
teppohudsson wants to merge 13 commits intomainfrom
feat/simplified-flow

teppohudsson commented Apr 28, 2026 •

edited

Loading

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

chatgpt-codex-connector Bot Apr 28, 2026

Uh oh!

teppohudsson Apr 28, 2026

Uh oh!

chatgpt-codex-connector Bot Apr 28, 2026

Uh oh!

teppohudsson Apr 28, 2026

Uh oh!

chatgpt-codex-connector Bot Apr 28, 2026

Uh oh!

teppohudsson Apr 28, 2026

Uh oh!

teppohudsson commented Apr 28, 2026 •

edited

Loading

Uh oh!

bmichele left a comment

Uh oh!

bmichele commented Apr 30, 2026

Uh oh!

bmichele left a comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

teppohudsson commented Apr 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

1. User journey analysis (user-journey.yaml / TTV)

2. Push: entire Skene bundle directory (e.g. skene-context/)

Test plan

Checklist

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector Bot Apr 28, 2026

Choose a reason for hiding this comment

Uh oh!

teppohudsson Apr 28, 2026

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector Bot Apr 28, 2026

Choose a reason for hiding this comment

Uh oh!

teppohudsson Apr 28, 2026

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector Bot Apr 28, 2026

Choose a reason for hiding this comment

Uh oh!

teppohudsson Apr 28, 2026

Choose a reason for hiding this comment

Uh oh!

teppohudsson commented Apr 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

bmichele left a comment

Choose a reason for hiding this comment

Uh oh!

bmichele commented Apr 30, 2026

Uh oh!

bmichele left a comment

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

teppohudsson commented Apr 28, 2026 •

edited

Loading

1. User journey analysis (`user-journey.yaml` / TTV)

2. Push: entire Skene bundle directory (e.g. `skene-context/`)

teppohudsson commented Apr 28, 2026 •

edited

Loading