AgentWorkforce/burn relayburn-v3.3.0 on GitHub

Packages

relayburn@3.3.0 (tag: relayburn-v3.3.0)
@relayburn/sdk@3.3.0 (tag: sdk-v3.3.0)
@relayburn/mcp@3.3.0 (tag: mcp-v3.3.0)
@relayburn/cli-darwin-arm64@3.3.0 (tag: cli-darwin-arm64-v3.3.0)
@relayburn/cli-darwin-x64@3.3.0 (tag: cli-darwin-x64-v3.3.0)
@relayburn/cli-linux-arm64-gnu@3.3.0 (tag: cli-linux-arm64-gnu-v3.3.0)
@relayburn/cli-linux-x64-gnu@3.3.0 (tag: cli-linux-x64-gnu-v3.3.0)
@relayburn/sdk-darwin-arm64@3.3.0 (tag: sdk-darwin-arm64-v3.3.0)
@relayburn/sdk-darwin-x64@3.3.0 (tag: sdk-darwin-x64-v3.3.0)
@relayburn/sdk-linux-arm64-gnu@3.3.0 (tag: sdk-linux-arm64-gnu-v3.3.0)
@relayburn/sdk-linux-x64-gnu@3.3.0 (tag: sdk-linux-x64-gnu-v3.3.0)

relayburn-sdk@3.3.0 (tag: relayburn-sdk-v3.3.0) — https://crates.io/crates/relayburn-sdk/3.3.0
relayburn-cli@3.3.0 (tag: relayburn-cli-v3.3.0) — https://crates.io/crates/relayburn-cli/3.3.0

burn summary and burn hotspots no longer run a pre-query ingest sweep by default — both return in well under 100ms instead of seconds on large ledgers (hotspots was ~3s). Pass --ingest for a one-off freshen; keep the ledger current out of band with burn ingest --watch or the Claude Stop hook.
Faster reads on large ledgers: time-windowed turn queries (--since/--until without a --session) now seek the ts index instead of scanning the whole turns table, cutting summary/hotspots/overhead/compare query time by roughly 3-4x. Output is unchanged.