Infrastructure & Operational Memory for Modern Engineering Teams
Preserve operational continuity as infrastructure evolves.
OpsLocker is a local-first operational memory platform. It helps engineering teams remember what they run, who owns it, what it costs, what depends on what, and what's quietly bleeding money — without forcing them to upload that map into someone else's cloud.
Built by Mpingo Systems LLC — Eddy Mkwambe, Technical Director.
npm install
npm run db:generate
npm run db:migrate
npm run db:seed # optional — populates a realistic demo workspace
npm run devApp runs at http://localhost:3000. On a fresh database you'll be sent to /setup to create your first workspace; with the seed loaded you'll land on /dashboard.
A persistent operational memory layer for the infrastructure your team relies on.
It answers questions your tools currently can't:
- Why does this service exist?
- Who owns this vendor?
- Is this API still being used?
- When does this renew?
- Why are we still paying for this?
- What depends on this if we turn it off?
It is not a password manager, a SIEM, a cloud observability platform, an accounting suite, or runtime monitoring. Secrets are optional supporting metadata, not the product's identity.
Resources are organized into projects inside a workspace. Each resource has a category, lifecycle state, owner, environment, monthly cost, renewal date, and operational notes.
Categories:
- Database — Postgres, SQLite, managed datastores
- API — third-party APIs (OpenAI, Anthropic, Stripe, …)
- Domain — custom domains and DNS providers
- Cloud — hosting platforms, IaaS, PaaS
- Auth — identity providers (Clerk, Auth0, Cognito, …)
- CI/CD — source control and pipelines
- Analytics — product analytics, BI
- Communication — email, Slack-style, transactional
- Storage — blob, object, file storage
- Monitoring — APM, log aggregation, uptime
- Subscription — pure billing relationships
- Other — anything that doesn't fit cleanly
Lifecycle states (govern what surfaces in dashboards and risk panels):
active·trial·at_risk·deprecated·archived
Relationships map operational dependencies between resources:
depends_on·bills_through·authenticates_with·sends_through·deploys_to·integrates_with
Every change — adding a resource, rotating a secret, flagging a renewal, acknowledging a reminder — is written to an immutable operational timeline.
- Dashboard — operational overview, renewal risk, recent activity, waste signals
- Projects — group infrastructure by application or environment
- Registry — full resource inventory with filters, sortable table, and a slide-over detail drawer
- Graph — React Flow dependency map with dagre auto-layout and orphan detection
- Timeline — chronological event feed grouped by date with type filters
- Subscriptions — renewal calendar, lifecycle breakdown, inactive vendor review
- Finances — vendor spend table, spend by category/project, monthly trend, waste signals, invoices
- Reminders — severity-grouped reminders with optimistic acknowledge
- Search — debounced full-text search across vendors, owners, notes, tags
- Settings — workspace identity, data ownership, import/export
OpsLocker is most useful with your real inventory loaded — and most users don't want to type it in by hand. Two import paths:
Settings → Data → Import resources → CSV tab, or hit the Import button on the registry empty state.
Expected columns (extras are ignored, missing columns are treated as empty):
name, vendorName, category, environment, owner, lifecycleState,
renewalDate, monthlyCost, currency, notes, tags, website
Each row is validated with Zod; invalid rows are skipped and reported with the specific failure reason. There's a Download template button in the dialog that generates a sample CSV with the correct headers and one example row.
Settings → Data → Import resources → .env file tab.
Paste an .env file content and OpsLocker will create one resource per KEY=value pair (category: api, lifecycle: active). Comments, blank lines, malformed lines, and duplicate keys are skipped — counts are shown in the live preview before you commit.
CRLF line endings are handled.
- ⌘K / Ctrl+K — open the command palette. Jump to any page, create a resource, create a project, or open the importer in a single keystroke. The palette also live-filters resources by name.
OpsLocker stores everything in a SQLite file on your machine (opslocker.db in the project directory by default; override with DATABASE_PATH). There is no required cloud sync, no third-party data dependency, and no vendor lock-in.
This is intentional. Infrastructure maps, vendor relationships, billing metadata, and operational notes are exactly the kind of data engineering teams have learned to be careful about handing to centralized services. SQLite is portable, inspectable, backup-friendly, and outlasts any specific tool — including OpsLocker.
Export is first-class. From Settings → Data → Preview & export you get a per-table count of exactly what's about to leave, then a workspace-named JSON or CSV download:
opslocker-{workspace-slug}-{YYYY-MM-DD}.json
opslocker-{workspace-slug}-{YYYY-MM-DD}.csv
If you ever want to migrate, host this yourself, or just walk away with your data, the export is the whole story.
- Next.js 15 (App Router) · React 18 · TypeScript strict
- SQLite via
better-sqlite3· Drizzle ORM with typed migrations - Tailwind CSS 3 · shadcn/ui · Lucide icons · dark theme only
- TanStack Table for tables · React Flow + dagre for the dependency graph · Recharts for finance charts
- React Hook Form + Zod for every form · Zustand for global UI state
- cmdk for the command palette · Sonner for toasts · PapaParse for CSV
Run npm run db:seed to load demo data before capturing screenshots, then save them under screenshots/ at the repo root:
screenshots/dashboard.pngscreenshots/projects.pngscreenshots/registry.pngscreenshots/graph.pngscreenshots/timeline.pngscreenshots/subscriptions.pngscreenshots/reminders.png
docs/OpsLocker-PRD-and-TRD.md— product and technical requirementsdocs/OpsLocker-Value-Proposition-and-Monetization.md— positioning and business modeldocs/CLAUDE-CODE-EXECUTION-PROMPT.md— original sprint plan
Pre-release (0.1.0). Local-only. The data model, API surface, and route structure are stable enough to build against; future sprints may add optional encrypted sync, multi-user collaboration, and integrations.