Methodology

What Fusion Monitor checks

Fusion Monitor tracks how IPOR, Fusion and selected Fusion vaults appear across external DeFi platforms. It checks names, descriptions, approved assets, public links, listings and selected vault metadata. It does not assess protocol safety or update external platforms directly.

Reference and evidence

• Reference: approved names, links, assets and entity identities, including relevant vault information from the Fusion API.
• Observation: snapshots of information exposed by each third-party source, retained with provenance.
• Action records: stable findings, review decisions and recorded status transitions used to coordinate follow-up.

Reference information, collected observations and human decisions remain distinct. A comparison indicates a discrepancy to assess, not an instruction to change the approved reference.

Seven-stage workflow

1. Reference. Establish the approved value or identity against which the source will be checked.
2. Observation. Collect the source’s exposed fields and preserve their evidence.
3. Comparison. Compare observed values with the reference and previous successful snapshot.
4. Finding. Record a discrepancy under a stable identity, with priority and supporting evidence.
5. Action. Review the finding, take responsibility and record follow-up or an accepted limitation.
6. Refresh. Collect the source again after an external update. An absent discrepancy becomes a possible correction.
7. Verified resolution. A reviewer checks the evidence and records the finding as resolved. A successful collection alone does not verify resolution.

Collection and freshness

Routine refreshes use source-specific adapters and deterministic extraction. External sources are refreshed when authorised team members need to inspect or verify them.

• A successful complete collection produces a new snapshot.
• Failed or partial collection preserves the latest reliable successful state.
• Collection failures are distinct from representation findings.
• Ageing or stale evidence should be refreshed before relying on it for follow-up.

Comparison and confidence

Observations retain source, entity, field, value, timestamp and evidence references. The new snapshot is compared with the captured Fusion reference and the previous successful source snapshot. Recorded changes retain before and after values.

Confidence describes the reliability of the collected evidence. High confidence indicates a direct deterministic source; medium or low confidence calls for additional checking. Confidence, priority and collector health describe different things.

Findings and review

A recurring discrepancy retains its finding identity. Use Review to inspect the expected and observed values, update task status and record a review note.

• Saving a non-open status assigns the finding to the acting reviewer. Setting it to Open clears assignment.
• In progress and Awaiting external update record the follow-up stage.
• Possible correction means the discrepancy was absent from a complete refresh and still needs reviewer verification.
• Resolving records the reviewer’s decision. Changed evidence can reopen a finding or require another review.
• Recorded status transitions, actors and timestamps are available in History. Earlier assignment changes are not separately retained as a complete assignment log.

Limits

• External sources expose different fields and update at different times.
• Collection availability and representation quality are separate states.
• Ambiguous changes require human confirmation.
• Fusion Monitor is read-only relative to Fusion and external platforms, and never signs protocol transactions.
• Viewer users cannot change findings; authorised roles manage the internal review workflow.
• The Demo uses fictional data and exposes no operational notes or configuration.