docs: add a three-track user manual and absorb the existing user guide (#428) - #429
docs: add a three-track user manual and absorb the existing user guide (#428)#429w7-mgfcode wants to merge 2 commits into
Conversation
…428) Add docs/manual/ — 23 chapters across an operator, analyst, and integrator track, plus shared configuration, troubleshooting, FAQ, and glossary references. Each chapter carries a Purpose / Intended reader pair, a "What you'll accomplish" section, and a Next pointer. Stated numbers trace to app/core/config.py, a fixed constant, or an authoritative module; runtime figures are explained rather than asserted. Corrects two drifts found against the code while writing: - the model list was 7 types; app/shared/model_taxonomy.py defines 11 - the data-platform table count was 7; there are 10 Six user-guide files become redirect stubs. showcase-manual-demo-guide.md (reviewer QA procedure) and showcase-walkthrough.md (unshipped roadmap) are kept intact with a pointer banner — neither is user-manual content. Every internal link and anchor resolves; no chapter is orphaned.
There was a problem hiding this comment.
Sorry @w7-mgfcode, your pull request is larger than the review limit of 150000 diff characters
|
Important Review skippedAuto reviews are disabled on base/target branches other than the default branch. Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Plus Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
Reviewer's GuideAdds a new three-track user manual under docs/manual/, migrates and redirects the previous user-guide content into it, and corrects documented model and data-platform counts against authoritative code sources, leaving two non-manual guides intact with pointer banners. File-Level Changes
Possibly linked issues
Tips and commandsInteracting with Sourcery
Customizing Your ExperienceAccess your dashboard to:
Getting Help
|
Ran the stack end to end and checked the quickstart chapter against real
output. Four corrections:
- /health returns {"status":"ok","database":null}, not {"status":"ok"}.
Fixed in installation, troubleshooting, and api-reference. (The root
README carries the same inaccuracy.)
- `docker compose up -d` starts postgres, backend AND frontend — only
ollama is profile-gated. Host-mode instructions now name the service:
`docker compose up -d postgres`.
- The demo trains three baselines (naive, seasonal_naive,
moving_average), so its winner says nothing about feature-aware
models. Removed the misleading implication and pointed at the champion
selector for real comparisons.
- reset/agent/cleanup legitimately report skip and the run still exits 0
GREEN. Documented skips as non-failures.
Adds a troubleshooting entry for "Multiple head revisions are present",
which breaks `alembic upgrade head`, the backend container's startup
command, and `make demo`.
Verified as correct: exit code 2 on precondition failure, the 11-step
pipeline order, the summary-line format, SHA-256 artifact verification,
graceful agent skip without an API key, and seed reproducibility — a
re-run on seed 42 reproduced the row count, date range, selected pair,
and WAPE to four decimals.
Closes #428.
Adds
docs/manual/— a navigable user manual organised into three tracks — and folds the existingdocs/user-guide/into it.Why three tracks
IntentGuard's manual uses two (operator / integrator). ForecastLabAI needs a third: it has a dashboard-analyst persona who never opens a terminal, and folding those chapters into the operator track buries them.
23 chapters, ~26k words.
Conventions
Every chapter carries a
Purpose:/Intended reader:header pair, a "What you'll accomplish" section, dense cross-linking, and a "Next" pointer — so any chapter is a valid entry point.Evidence discipline. Stated numbers trace to
app/core/config.py, a fixed constant, or an authoritative module (app/shared/model_taxonomy.py,app/shared/feature_frames/contract_v2.py). Runtime-dependent figures — durations, accuracy, which model wins — are explained rather than asserted, since they depend on data, seed, and hardware. The synthetic-data caveat is stated once inoperator/concepts.mdand referenced rather than repeated.Drift corrected
Found by checking the old guide against the code:
feature-reference.mdlisted 7;_MODEL_FAMILY_MAPdefines 11 — 5 baseline, 4 tree, 2 additive. (weighted_moving_average,seasonal_average,random_forest,trend_regression_baselinewere missing.)exogenous_signal,replenishment_event,sales_returns).Same class of drift as #376 (stale slice count in
docs/_base/). The manual states 19 slices and 23 tables, both verified.Disposition of docs/user-guide/
Six guides → short redirect stubs pointing at their new chapters.
Two are kept intact with only a pointer banner added, because neither is user-manual content:
showcase-manual-demo-guide.md— a reviewer QA procedure.showcase-walkthrough.md— contains an unshippedPlanned (PRP-38..41)roadmap.Validation
docs/manual/anddocs/user-guide/resolves to an existing file.Docs-only change — no application code touched, so the lint/type/test gates are unaffected.
Summary by Sourcery
Introduce a new three-track user manual under docs/manual and fold the existing user-guide content into it with validated, up-to-date information on models, tables, configuration, operations, and workflows.
Documentation: