Vendor profile

Anthropic / Claude / Claude Code

Claude chat, Anthropic API, Claude Code, and related developer surfaces

No v0.2 run receipt. Scores remain TBD. Coverage confidence: medium. Source label: Confirmed by vendor.

Status

Coverage statusQueued for v0.2 measurement
Correction channelSupport channel documented; vendor response intake not exercised.
NotesPublic incidents can trigger tests, but scores wait for 121 run receipts.

Public source links

LinkSource labelConfidence
HomepageConfirmed by vendormedium
API docsConfirmed by vendormedium
PricingConfirmed by vendormedium
Trust centerConfirmed by vendormedium
Privacy policyConfirmed by vendormedium
SupportConfirmed by vendormedium

Score state

CategoryLabelConfidenceSourceNote
Model identity disclosureTBDnot-yet-measuredUnknownNo v0.2 Anthropic run receipt yet.
Reroute and fallback visibilityTBDnot-yet-measuredUnknownNo v0.2 Anthropic run receipt yet.
Silent capability driftTBDnot-yet-measuredUnknownNo v0.2 Anthropic run receipt yet.
System prompt and behavior-change disclosureTBDnot-yet-measuredUnknownNo v0.2 Anthropic run receipt yet.
Context preservation integrityTBDnot-yet-measuredUnknownNo v0.2 Anthropic run receipt yet.
Data retention and human access clarityTBDnot-yet-measuredUnknownNo v0.2 Anthropic run receipt yet.
Pricing, quota, and effort clarityTBDnot-yet-measuredUnknownNo v0.2 Anthropic run receipt yet.
Internal and external capability parity disclosureTBDnot-yet-measuredUnknownNo v0.2 Anthropic run receipt yet.
Tool use and source provenanceTBDnot-yet-measuredUnknownNo v0.2 Anthropic run receipt yet.
Correction and redress pathTBDnot-yet-measuredUnknownNo v0.2 Anthropic run receipt yet.
Safety boundary clarityTBDnot-yet-measuredUnknownNo v0.2 Anthropic run receipt yet.
Machine-readable trust receiptTBDnot-yet-measuredUnknownNo v0.2 Anthropic run receipt yet.
Substrate continuityTBDnot-yet-measuredUnknownNo v0.2 Anthropic run receipt yet.

Watchpoints

  • Visible fallback and policy-route disclosure.
  • Differences between chat, API, and code-agent surfaces.
  • How correction history is preserved after public incident response.