Detailed Analysis
A solo developer has released two free third-party tools — Ogma and Tyr — designed to address a significant operational gap in Anthropic's current tooling: the absence of a unified usage and audit dashboard spanning all Claude surfaces. Teams using Claude on Anthropic's Team plan currently face a fragmented visibility problem, with API key usage tracked through Anthropic's developer console, claude.ai project activity tracked separately, and Claude Code sessions logged in yet another location. Ogma consolidates these streams into a single per-user timeline, enabling administrators to see the complete Claude "footprint" of any individual employee across API, Claude Code, and Claude Desktop/web/mobile interactions within a single interface. The product is offered free during its preview period with no payment information required.
The most technically novel component of the offering is an MCP (Model Context Protocol) connector that enables logging of Claude Desktop, web, and mobile sessions — surfaces that have historically been invisible to enterprise administrators. Rather than attempting to silently intercept conversations, the connector logs a one-line summary of each turn as a visible tool call within the conversation itself, a design choice the developer explicitly frames as a response to Claude's own safety training, which refuses covert third-party data exfiltration. This transparency-by-design approach reflects a meaningful constraint: the tool captures Claude's own summaries of interactions, not raw prompt and response content, making it suitable for cost analytics, compliance metadata, and topic classification, but not forensic transcript purposes. Full content capture remains gated behind Anthropic's Enterprise plan and its Compliance API.
The audit infrastructure built around Ogma speaks to growing demand in regulated industries for defensible records of AI system usage. The product implements hash-chained event logs anchored daily to a public blockchain, creating tamper-evident audit trails that are architecturally independent of Anthropic's own infrastructure. The companion product, Tyr, extends this compliance posture to autonomous agent deployments, acting as an inline proxy that evaluates agent tool calls against Open Policy Agent (OPA) Rego policies and similarly hash-chains decisions to blockchain. Both products target an emerging category of enterprise need: the ability to prove, to a regulator or auditor, not just what an AI system was instructed to do, but what it actually did and when.
The project highlights a broader structural tension in the current enterprise AI landscape. Anthropic, like other frontier AI providers, has built distinct products — API access, consumer chat interfaces, developer tooling — that have accumulated separate administrative surfaces without a unified cross-product management layer. For individual users or small teams, this fragmentation is a minor inconvenience; for compliance-oriented enterprises, it represents a genuine audit gap. The fact that a solo developer was able to identify and partially address this gap using Anthropic's own MCP standard and Admin API illustrates both the extensibility of Anthropic's platform and the degree to which enterprise-grade administrative tooling remains underdeveloped relative to the underlying capabilities being offered.
The emergence of tools like Ogma reflects a pattern consistently visible across enterprise software cycles: foundational vendors build powerful capabilities while third-party developers, often working in regulated verticals, build the governance and observability layers the vendor has not yet prioritized. Anthropic's own trajectory — adding an Admin API, a Compliance API gated to Enterprise, and MCP support — suggests the company is aware of these needs, but the pace of that development has created a market for independent solutions. Whether Anthropic eventually absorbs this functionality natively, as has happened in analogous markets like cloud infrastructure, will depend partly on how much enterprise adoption pressure accumulates around exactly the compliance and cross-surface visibility pain points this developer has moved to address.
Read original article →