The agent assembled a partial but substantive explanation of Arc Studio by working around a direct 403 block on studio.arc.io, retrieving documentation and marketing materials from arc.io and docs.arc.io instead. Pricing and audience definition remained incomplete due to gaps in the published materials rather than navigation failures.

• ›The studio.arc.io homepage itself was inaccessible (HTTP 403), forcing the agent to rely on marketing copy from arc.io and technical docs — a significant friction point that the agent worked around without user intervention.
• ›Arc's docs explicitly mention Arc Studio in the llms.txt index and carry a dedicated page (arc-studio.md), making it discoverable to AI agents; the docs confirm testnet-only deployment and live preview capabilities, but omit pricing, gas costs, and mainnet migration paths.
• ›The agent correctly identified a content gap: pricing is stated nowhere in either the parent marketing site (which lists the platform as '0 USD' in structured data but provides no fee schedule) or the docs; the claim that testnet is free comes from inference (built-in faucet) rather than explicit policy.
• ›The parent site (arc.io) and docs describe Studio's audience differently — marketing frames it as 'describe an idea, get full-stack apps,' while docs present it as an 'AI coding agent' — suggesting inconsistent positioning or intentional scope ambiguity.
• ›The site embeds 'Agent Instructions' blocks in docs pages (telling AI agents to install skills and configure USDC), which the agent appropriately filtered out as not part of the product description, indicating awareness of instructional pollution.