Shared memory for scrum masters
Scrum masters carry the memory of how a team actually works: the impediments that keep recurring, the action items a retrospective produced, the dependency that bit the team last sprint, the reason a story was split or a commitment was missed. As teams adopt AI agents to summarize standups, draft retro notes, and track work, each agent starts blind to that history, so retro actions evaporate, the same impediment surfaces sprint after sprint, and the lessons a team agreed on never stick. Glen, shared memory for AI agents, gives your team's agents one durable, shared memory exposed as a single MCP tool, so the process history you steward is recalled instead of repeatedly rediscovered.
A scrum master's value is continuous improvement, and improvement depends on memory: a retro is only worth running if its actions are remembered and an impediment is only worth raising if the team learns from it. AI agents undercut that when they start cold, an agent drafting this sprint's retro does not know last sprint's actions were never closed, or that the same blocker has appeared three times, so the team relitigates problems it already named. Glen makes that process memory durable: connected over MCP, an agent reads the team's accumulated history before it summarizes or plans, picking up recurring impediments and open retro actions, and writes back new ones, so the team's improvement loop actually compounds.
For a scrum master this means the patterns you track become memory the agents consult rather than something you alone have to remember and re-raise. The recurring impediment, the unfinished action item, the dependency that caused a slip, all become context an agent recalls when it drafts a retro, plans a sprint, or summarizes a standup, so the team confronts its real patterns instead of cycling through them. Because Glen is org-scoped, this history is shared across the whole team's agents and beyond, so an impediment one agent records is visible to the agent helping plan and to a manager's agent looking across teams. As a standard MCP server, Glen complements the board and tracking tools your agents already touch, holding the durable process learnings none of them capture, and is readable from Claude Code, Cursor, or any MCP client. Connect once over OAuth or an API key and let the team's improvement compound.
FAQ
- Can it keep retrospective action items from evaporating?
- Yes. Agents record retro actions and recurring impediments as durable observations and read them back next sprint, so the team's improvement loop persists instead of resetting each retro.
- Does the whole team's agents share this history?
- Yes. Glen is org-scoped, so an impediment or action one agent records is readable by every teammate's agent, keeping the team aligned on its real patterns.