Plans and usage

Lore charges paid self-service plans per seat. A seat is a team member covered by the subscription. MCP reads are pooled across the organization: adding seats does not multiply the included read quota.

Included usage

Plan	Seats	Repositories	Shared monthly MCP reads
Free	1	5 active repositories	10,000
Team	Paid per seat	Unlimited	1,000,000
Business	Paid per seat	Unlimited	10,000,000
Enterprise	Agreed in your contract	Agreed in your contract	Agreed in your contract

See current pricing for monthly and annual prices. Billing in the dashboard shows your organization's plan and usage.

What counts as a read?

Each successful get_context, search, session_briefing, retrospective, or related_repositories call counts as one read, including calls that return no facts or use a cached result. The count is not based on the number of facts returned.

For example:

• One search returning 20 facts uses one read.
• Ten separate searches use ten reads.
• One get_context call with a larger response budget still uses one read.
• list_repos, diagnose_connection, record, and forget do not consume this read quota.
• A failed metered call refunds its reserved read.

An agent may make several calls to answer one prompt. One message to your agent therefore does not always equal one Lore read.

How does sharing work?

Suppose a Team organization has five members. All five share the same one million monthly reads. If one member uses 200,000 and the others together use 300,000, 500,000 remain. The allowance is not one million for each member.

Business works the same way with ten million shared reads.

When does usage reset?

Reads reset on the first day of each calendar month at 00:00 UTC. That reset is separate from the date Stripe charges your subscription. Annual billing does not turn the allowance into a single yearly pool; reads still reset monthly.

What happens at the limit?

Lore uses the included monthly allowance first, then purchased invocation credits. One credit pays for one additional successful metered call. If both are exhausted, an administrator can buy a configured pack in Settings → Plans & billing, upgrade, or wait for the monthly reset. No automatic overage charge or automatic refill occurs. Purchases are shared by the workspace, not multiplied by seats. Failed calls return their reserved credit. Credits do not change plan features or ingestion speed.

Pack quantity, price, and any expiration appear before checkout. Non-expiring packs carry forward across monthly resets. Payment must be confirmed before credits become available. Refunded purchases reduce credits proportionally; disputed purchases are unavailable while the dispute is unresolved. Credit packs may not yet be configured for your deployment. GitHub mention answers also consume one metered invocation.

What changes with processing priority?

Importing a repository and reading its existing memory are different operations. Scans use a processing queue: Free has low priority, Team has standard priority, and Business has higher priority and more concurrent processing capacity.

Unlimited repositories on a paid plan does not mean every repository is scanned simultaneously. A large first import can take time while jobs are processed.

Team processes one ingestion job at a time per workspace, with at least 60 seconds between starts. Repository scans and external-source syncs share that limit. Business allows up to three simultaneous jobs with higher repository queue priority and no start interval. Actual concurrency also depends on available worker capacity. Free uses one job at a time with a 15-minute start interval. These controls do not impose repository caps on paid plans or a monthly data-import quota.