FinOps, retry waste, self-hosted observability, and fleet visibility for teams running multiple AI coding agents.
A release that fixes badly inflated retry_count and token-usage numbers for real Claude Code CLI sessions, plus a backfill tool for data ingested under the old logic.
Read article: RetrySight Lite 1.1.0: accurate retry counts for Claude CodeSame retry taxonomy, three deployment models. A practical decision framework for solo developers, growing teams, and platform orgs choosing where their agent telemetry lives.
Read article: Lite, Cloud, or Enterprise: a field guide to running RetrySightRetrySight exposes retry and cost data over MCP, so Cursor, VS Code, or your Cloud MCP client can query fleet history directly instead of you switching to a dashboard.
Read article: Ask your coding agent about its own retriesEdit loops, rejected diffs, compaction, and subagent spirals all look like "tokens used" in a vendor dashboard. RetrySight classifies them separately — here's why that split matters.
Read article: What actually counts as a retry? RetrySight's classification, explainedThe managed SaaS portal is open at cloud-staging.retrysight.com — a hosted console for fleet retries and token spend.
Read article: RetrySight Cloud is now in early accessEngineering teams run Cursor, Claude Code, and Copilot side by side. Fleet visibility requires a vendor-neutral collector layer.
Read article: Why one vendor dashboard is not enough for multi-tool teamsWhy platform and security teams keep agent telemetry inside their network — and what a self-hosted deployment looks like.
Read article: Self-hosted observability for AI coding agentsRetry loops, compaction, and rejected diffs burn tokens without showing up in vendor dashboards — and what platform teams can do about it.
Read article: The hidden cost of coding agent retries
RetrySight