PROJECT GRAPH FEATURE
Inspecting the Graph
Once Project Graph is enabled, structural questions stop being archaeology. You can ask for exactly the slice, dependency, hotspot, or module view you need.
Feature Guide
A quick orientation block that answers the essential questions: what this feature does, how it works, why it matters, and the key concepts behind it.
What this does
Once the package is enabled and the graph exists, inspection becomes a set of explicit structural views instead of improvised archaeology. You can render slices, query dependencies, inspect whole modules, and ask the intelligence layer for hotspots, doc gaps, or event lifecycles.
How it works
Use `ai:review-graph:show` for readable slices, `ai:review-graph:query` for targeted dependency questions, `ai:review-graph:module` for a module-level overview, `ai:review-graph:intelligence` for higher-level structural explanations, and `ai:review-graph:context` when the task needs a review- or AI-ready context package.
Why it matters
This is where the package becomes operationally useful. Reviews get faster, onboarding becomes less fragile, and AI tools can start from architecture-backed answers instead of broad guesses assembled from random files.
Key concepts
- ai:review-graph:show
- Renders summary, markdown, JSON, or DOT graph views for a chosen focus, module, or node type.
- ai:review-graph:query
- Runs structural lookups such as search, usages, dependencies, and cross-module edge inspection.
- ai:review-graph:module
- Builds a module overview with summary counts, domain context, hotspots, and optional flows or event details.
- ai:review-graph:intelligence
- Queries the higher-level intelligence layer for hotspots, documentation gaps, flows, event lifecycles, intent, and natural-language structural answers.
- ai:review-graph:context
- Builds task-scoped structural context for review, refactor, debugging, or AI-assisted work.
Inspection Flow
Move from “what is in this repo?” to a focused structural answer
Inspection is not one command. It is a small family of surfaces that let you choose the right answer shape for the question: rendered slice, query result, module overview, or intelligence-backed explanation.
Step 1
Render the slice you need
Start broad when you need a structural snapshot, then narrow by module, node type, or focus target when the question is localized.
bin/semitexa ai:review-graph:show --format=markdown --module=Demobin/semitexa ai:review-graph:show --format=json --type=service,handlerbin/semitexa ai:review-graph:show --format=dot DemoStep 2
Ask focused structural questions
When the question is about coupling, usage, or module boundaries, query directly instead of reconstructing the answer from imports and memory.
bin/semitexa ai:review-graph:query --search=DemoCatalogServicebin/semitexa ai:review-graph:query --dependencies=Semitexa\\Demo\\Application\\Service\\DemoCatalogServicebin/semitexa ai:review-graph:query --cross-module --from=Demo --to=CoreStep 3
Reach for intelligence when edges are not enough
Some questions need richer answers than a node list: hotspots, event lifecycles, module summaries, inferred intent, and task context.
bin/semitexa ai:review-graph:module Demo --include-events --include-flows --format=jsonbin/semitexa ai:review-graph:intelligence --hotspotsbin/semitexa ai:review-graph:context "review Demo module coupling" --format=jsonPractical Benefit
- Reviewers can spot module leaks and hotspots from explicit graph-backed answers instead of architectural intuition alone.
- Onboarding gets easier because new engineers can ask structural questions without already knowing where to look.
- AI workflows improve because module and context views are narrower and more defensible than arbitrary file dumps.
- Teams save time because common dependency and module questions stop triggering another round of grep and tab-hopping.
Verified against Semitexa Ultimate 2026.09.19.1020
Inspecting the Graph
Once the graph exists, Project Graph becomes an explicit inspection surface rather than a one-off artifact.
Canonical flow
- Render the slice that matches the question.
- Query dependencies, usages, or cross-module edges directly.
- Reach for module and intelligence views when raw edges are not enough.
Commands
bin/semitexa ai:review-graph:show Demo --format=markdown
bin/semitexa ai:review-graph:query --search=DemoCatalogService
bin/semitexa ai:review-graph:query --dependencies=Semitexa\\Demo\\Application\\Service\\DemoCatalogService
bin/semitexa ai:review-graph:module Demo --include-events --include-flows --format=json
bin/semitexa ai:review-graph:intelligence --hotspots
Why this matters
Architectural questions become focused graph-backed answers instead of archaeology. That improves onboarding, review speed, and the quality of AI context.
How it works
Use `ai:review-graph:show` for readable slices, `ai:review-graph:query` for targeted dependency questions, `ai:review-graph:module` for a module-level overview, `ai:review-graph:intelligence` for higher-level structural explanations, and `ai:review-graph:context` when the task needs a review- or AI-ready context package.
Why it matters
This is where the package becomes operationally useful. Reviews get faster, onboarding becomes less fragile, and AI tools can start from architecture-backed answers instead of broad guesses assembled from random files.
Key concepts
- ai:review-graph:show
- Renders summary, markdown, JSON, or DOT graph views for a chosen focus, module, or node type.
- ai:review-graph:query
- Runs structural lookups such as search, usages, dependencies, and cross-module edge inspection.
- ai:review-graph:module
- Builds a module overview with summary counts, domain context, hotspots, and optional flows or event details.
- ai:review-graph:intelligence
- Queries the higher-level intelligence layer for hotspots, documentation gaps, flows, event lifecycles, intent, and natural-language structural answers.
- ai:review-graph:context
- Builds task-scoped structural context for review, refactor, debugging, or AI-assisted work.
When This Shines
High-friction situations that become much easier
The graph is especially valuable when normal repository familiarity breaks down.
Start with `show` or `module` when you need a readable slice, then drop to `query` only when the question is specific.
Use cross-module and dependency inspection before “small” cleanup tasks that might secretly cross boundaries.
Use JSON or Markdown output when the result needs to be shared, reviewed, or consumed by automation.
Use `intelligence` or `context` when the goal is explanation or task preparation rather than raw graph edges.