Context & problem
A proposed inventory capability for enterprise application ecosystems with mixed languages, delivery tools, and runtime environments.
Problem
Modernization decisions require a reliable inventory of applications and their dependencies. Repository names alone do not establish runtime, ownership, lifecycle support, or migration readiness.
My contribution
I am outlining the inventory model, discovery boundaries, and review workflow. The scanner, integrations, and migration assessment remain planned work.
Architecture & decisions
Read-only discovery would collect repository and pipeline metadata. Classifiers would detect languages, manifests, package managers, build tools, and runtime declarations. A normalized inventory would retain provenance, observation time, confidence, and owner review.
Reference sequence: Read-only discovery → Classify with sources → Map dependencies → Review readiness.
Technology
Proposed mechanisms include Git metadata, manifest analysis, JSON inventory records, and pipeline configuration analysis. Implementation language and integrations are undecided.
Key decisions
Separate observed facts from inferred classifications. Preserve unknown values, record source revision, and require owner confirmation for runtime mappings and migration readiness.
Delivery & validation
Planned stages are read-only discovery, language and package-manager detection, dependency classification, pipeline/runtime mapping, and owner-reviewed modernization assessment. No working scanner or dashboard is available yet.
Security & operations
Use read-only discovery with explicit repository scope. Exclude source secrets and personal data; retain provenance and owner review for classifications. An inventory entry does not grant permission to modify the described service.
Validation
For a future scanner, compare language, runtime, build and ownership classifications with a manually reviewed sample. Include monorepos, shared pipelines, missing manifests and stale metadata. Report unknowns separately from incorrect classifications.
Trade-offs
Static manifests can be stale or incomplete. Polyglot repositories and shared pipelines complicate classification. Broader discovery increases access needs, so collection should minimize privileges and exclude secrets.
Outcomes & status
The intended output is a traceable inventory for modernization decisions. Coverage, classification accuracy, and migration value require evaluation; no measured results are available.
Scope & status
Planned design with no completed implementation. A working scanner and reviewed evaluation are the next milestones.