Appropriate use
Use for customer-reported device, browser, content, network, account, or environment problems.
Transforms a support case into a privacy-safe, controlled reproduction plan that compares customer conditions with known-good conditions and returns a clear confidence-based outcome.
This page defines the behaviour, controls, evidence, and outputs required from the skill. It does not claim that a language model can enforce security or correctness by wording alone.
Use for customer-reported device, browser, content, network, account, or environment problems.
Do not use to access a customer tenant, data, device, or network beyond explicit authorisation or to dismiss a report simply because the laboratory cannot reproduce it.
Each phase has explicit inputs and outputs. Approval gates are hard stops bound to exact versions and hashes.
Resolve the support case revision, authorised data access, customer consent, privacy restrictions, affected account, device, content, and communication owner.
Capture model, OS, firmware, app or player version, account state, role, region, network, proxy, DNS, content type, schedule, orientation, resolution, and recent changes.
Describe the observable symptom, timing, frequency, recovery, last-known-good state, error messages, screenshots, logs, and customer actions without inferring cause.
Create the smallest set of comparisons that isolates customer-specific variables from product, platform, network, content, and account effects.
Run controlled comparisons, preserve all attempts, avoid destructive account changes, and clearly separate laboratory evidence from customer-provided evidence.
Classify whether the signal follows content, device, firmware, network, account, version, configuration, time, or remains unknown.
Prepare a customer-safe conclusion, next diagnostic request, workaround when verified, and product defect proposal when evidence supports it.
Update the support case idempotently, link restricted evidence appropriately, and hand product defects or incidents into the correct workflow.
An approval cannot be reused after its environment, source, action plan, or approved document changes.
| Gate | Decision | Required binding |
|---|---|---|
| 1 | Gate 1 · approve case boundary | Approval stores approver, role, timestamp, context hash, approved document hash, expiry, and invalidation state. |
| 2 | Gate 2 · approve tests | Approval stores approver, role, timestamp, context hash, approved document hash, expiry, and invalidation state. |
| 3 | Gate 3 · confirm response | Approval stores approver, role, timestamp, context hash, approved document hash, expiry, and invalidation state. |
| 4 | Gate 4 · publish | Approval stores approver, role, timestamp, context hash, approved document hash, expiry, and invalidation state. |
Attempt outcomes remain separate from these workflow results. Errors, cancellations, and invalid context are never silently converted into a pass.
The customer failure signature was reproduced under matching controlled conditions.
A related symptom was reproduced, but one or more customer-specific aspects remain different.
Evidence links the issue to device, firmware, network, account, content, or configuration conditions.
The tested matrix did not reproduce the issue; this is not proof the report is invalid.
Critical environment or failure details are missing.
Evidence points to an external service or dependency.
Evidence has provenance, capture context, privacy classification, retention, and integrity metadata. A convenient artifact is not automatically sufficient proof.
| Claim | Primary evidence | Required validation |
|---|---|---|
| Customer symptom | Approved screenshots, video, logs, or case text | Privacy and provenance are recorded. |
| Environment | Device and configuration inventory | Unknown values remain unknown rather than guessed. |
| Comparison | Controlled matrix results | Only one or a justified set of variables changes per comparison. |
| Network | Approved connectivity and request evidence | Customer secrets and internal addresses are protected. |
| Content | Source, format, and known-good comparison | Customer content is handled according to consent and retention rules. |
| Outcome | Attempt and isolation summary | Confidence and unreproduced differences are explicit. |
These controls belong in the runtime, connector permissions, sandbox, renderer, storage layer, and review process.
No customer data or tenant access is used outside the recorded authorisation.
A lab success cannot erase a credible customer failure.
Prefer synthetic or customer-supplied minimal content over copying full production data.
Clear cache, reset, re-pair, or factory reset require impact-aware approval.
External responses avoid internal speculation, raw secrets, and blame.
Widespread or severe signals move to incident or security handling promptly.
The first applicable high-severity outcome takes precedence. A manual override is recorded separately and never rewrites the calculated result.
Request specific missing evidence or environment details.
Proceed to bug triage or incident flow based on impact.
Provide verified isolation steps and route to the correct owner.
Continue targeted isolation or file a qualified finding.
Report tested scope and next diagnostic option without dismissing the customer.
The workflow documents what it cannot prove and what further evidence would change confidence.
Human summaries are generated from validated structured records. They are not maintained as separate, drifting sources of truth.
case-context.jsonVersioned output produced by the workflow.
consent-record.mdVersioned output produced by the workflow.
environment-matrix.yamlVersioned output produced by the workflow.
failure-signature.mdVersioned output produced by the workflow.
reproduction-plan.mdVersioned output produced by the workflow.
attempt-ledger.jsonlVersioned output produced by the workflow.
isolation-analysis.mdVersioned output produced by the workflow.
customer-response.mdVersioned output produced by the workflow.
internal-handoff.mdVersioned output produced by the workflow.
customer-repro-report.htmlVersioned output produced by the workflow.
artifacts/qa/customer-issue-repro/<run-id>/ ├── case-context.json ├── consent-record.md ├── environment-matrix.yaml ├── failure-signature.md ├── reproduction-plan.md ├── attempt-ledger.jsonl ├── isolation-analysis.md ├── customer-response.md ├── internal-handoff.md └── customer-repro-report.html
Reproduce a customer report where Ethernet works but Wi-Fi does not on one SCOS device. Do not ask them to reset or clear cache yet.
Outcome: ENVIRONMENT-SPECIFIC Evidence: - Same device and content work over Ethernet. - Wi-Fi obtains an IP address but TLS requests fail. - A laboratory device on another Wi-Fi network works. - No cache or pairing reset was performed. Next customer-safe step: Test the affected device on a mobile hotspot and collect the exact time of the attempt. Internal direction: Investigate customer Wi-Fi proxy, firewall, DNS, or TLS interception before treating this as a player defect.
The written skill is only one layer. The repository, runtime, connectors, evidence storage, and publication path must implement these requirements.
Transforms a support case into a privacy-safe, controlled reproduction plan that compares customer conditions with known-good conditions and returns a clear confidence-based outcome.