Vestigivo RCA
Incident Intelligence & Root Cause Analysis for Visual Studio Code
Every incident leaves evidence.
A structured workspace for incident investigations — not a black-box root-cause oracle.
Actual Visual Studio Code screenshot · Vestigivo RCA 0.1.1 candidate · fully synthetic demonstration data
Vestigivo RCA brings together logs, distributed traces, exact-identifier correlations and investigator notes in a single Control Center inside the editor. It preserves the separation between what was observed, what was correlated, what is still a hypothesis and what an explicitly controlled check can establish.
Why Vestigivo?
Incidents often involve several services, timestamps with differing precision, incomplete spans and separate sources. The risk is not just missing a signal: it is mistaking temporal proximity or a plausible story for proof.
Vestigivo turns selected evidence into a traceable, bounded investigation. It records where each observation came from and explicitly marks missing or ambiguous information. It never performs automatic remediation, uploads local logs or claims that correlation proves causation.
Current capabilities (0.1.2)
| Area |
What it does |
| Evidence import |
Explicit local import of JSON, JSONL, NDJSON, structured .log and OpenTelemetry JSON logs/spans, with SHA-256 provenance and bounded parsing |
| Timeline |
UTC date filters, text, severity, service and exact trace/request-ID filtering; pages of up to 100 observations |
| Trace investigation |
Trace/span parent relationships; missing parents, cycles and ambiguous duplicate parents are identified rather than hidden |
| Correlation |
Exact trace ID, request ID and parent-span matching, always labeled CORRELATED — not causal proof |
| Case Studio |
Create, rename, duplicate, archive, annotate, bookmark and compare cases; local persistence is an explicit choice |
| Hypotheses |
Human-authored falsifiable statements and verification plans with assertions, preconditions, expected outcomes, risks and cost — no automatic execution |
| Verification |
Conservative evaluation contracts and distinct outcomes for introduced, pre-existing, absent, non-comparable or inconclusive observations |
| Reports |
Redacted JSON and Markdown export with limitations and sources; oversized exports fail rather than silently lose records |
| Git context |
Optional, read-only recent history in trusted workspaces; never automatically attributes a problem to a commit |
| External context |
Optional read-only Sentry issues and Grafana Cloud annotations, with per-read consent and tokens stored in VS Code SecretStorage |
| Interface |
Central editor Control Center with English/Spanish navigation, VS Code theme tokens, keyboard tab switching and no external scripts |
Control Center
Use the Vestigivo RCA icon in the Activity Bar or run Vestigivo RCA: Open Control Center. The main interface opens in an editor tab, not as a narrow sidebar.
Overview · Timeline · Evidence · Trace Tree · Correlations · Git Context · Report · Case Studio · Settings
To try a local investigation, choose Import evidence, select up to five supported files and inspect the resulting timeline and source fingerprints. In Case Studio, record a falsifiable hypothesis and — if appropriate — a manual verification plan. Save or export only when you explicitly choose to.
Explore the investigation workspace
All four screenshots below were captured from the actual Vestigivo RCA webview running inside the Visual Studio Code Extension Development Host on Linux, not redrawn or visually simulated. The incident is a clearly labeled synthetic fixture, parsed through the same evidence pipeline as local files. No real customer logs, tokens or incident data were used.
01 — Timeline: search, filter and inspect
Use exact trace/request identifiers, UTC date ranges, service and severity filters. Records retain source provenance and missing timestamps are not invented.

02 — Trace Tree: follow observed relationships
Explore span ancestry and inspect what is known, missing or ambiguous. Shared trace identifiers do not establish causation.

03 — Case Studio: build an evidence-backed investigation
Create and manage hypotheses, add notes, bookmark observations and document manual verification plans. A proposed explanation does not become a verified finding merely because it is recorded.

Capture provenance: media/screenshots/ · genuine Linux VS Code Electron/Xvfb screenshots, 1440 × 900, generated by the reproducible visual QA harness. The synthetic JSONL fixture is at test/fixtures/visual-demo.jsonl. See the technical checklist.
Evidence and confidence
| State |
Meaning |
OBSERVED |
A record was imported with provenance. It may still be incomplete or inaccurate. |
CORRELATED |
Records share exact identifiers. This does not show causality. |
HYPOTHESIS |
A possible explanation requiring a falsifiable check. |
CONFIRMED |
A narrow assertion passes the comparison and controlled-intervention gates; not a claim about the whole incident. |
INCONCLUSIVE |
The evidence or comparison is not sufficient to assert that specific conclusion. |
A verification plan with status PLANNED is not a test execution. Baseline/current comparisons distinguish introduced and pre-existing observations, never automatically prove a regression was caused by a particular change.
Privacy and trust model
- Local-first: no automatic file scanning, uploads, telemetry or AI calls by Vestigivo.
- Opt-in access: local imports use VS Code file pickers; Git inspection requires workspace trust and authorization; Sentry/Grafana require separate connection and an explicit read approval.
- Secret handling: provider credentials remain in VS Code SecretStorage. Reports and stored cases use best-effort redaction; no guarantee that every secret or personal datum is detected.
- Storage: cases are ephemeral until explicitly saved. Saved cases are not encrypted by Vestigivo; OS/VS Code protections apply.
- No remediation: never executes a captured command, changes a repository or deploys infrastructure to resolve an incident.
See Threat Model, Architecture and Acceptance Evidence.
JSON-based logs and OTLP JSON are supported. OTLP/gRPC Protobuf, a local OTLP collector, streaming imports, automated verification execution and broad causal inference are not implemented. Timestamp ambiguity remains explicit. Source limits default to 8 MiB per file and 10,000 observations per case.
Core tests, packaging and genuine Extension Host smoke now pass on Linux, Windows and macOS. WSL, SSH, Dev Containers, in-depth graphical accessibility audits and live cloud-provider checks remain release gates; do not infer certification from a passing unit test.
Development
Prerequisites: Node.js 22, npm and Visual Studio Code 1.96+.
npm ci --ignore-scripts --no-fund
npm run check # TypeScript, core tests and coverage gates
npm run brand:verify
npm run benchmark # Synthetic 1K / 10K / 100K budget exercise
npm run package:vsix # vestigivo-<version>.vsix
For a genuine Extension Development Host on a graphical environment, build first and run npm run test:host. On Linux CI, xvfb-run -a npm run test:vsix-host tests the installed VSIX bytes in an isolated profile.
Release policy: 0.1.0 was bootstrapped manually and 0.1.1 was published via Trusted Publishing. 0.1.2 is the next candidate, adding a GitHub Release alongside Marketplace and Open VSX. Pull requests execute validation; only an approved, version-bumping merge into main may initiate automated publishing. The repository visibility is controlled separately by the owner; publishing does not change it. See GitHub Releases, Release Readiness, Release Setup, Master Specification and Technical Checklist.
Language
English is used for the marketplace README and the main technical API names. Documentación en español.
License and authorship
Distributed under the MIT License. The repository is public; the software and contributions are covered by the terms in the license file.
Designed and developed by Alejandro Daniel Di Stefano
© 2026 ArmoTuSitio · Open source under the MIT License