# OrgComputers A shared workspace that gives AI coding agents common project context, memory, and tools. - Findings24: - ウェブサイト: - 料金: Freemium - カテゴリー: AIエージェント基盤, AIナレッジベース, MCPツール ## OrgComputersとは？ OrgComputers is a shared workspace for teams using AI coding agents. It combines project context, sessions, documents, code references, notes, links, connected tools, and activity so agents working in Claude Code, Cursor, Codex, or another MCP-compatible client can use the same team knowledge. Its core purpose is to help an agent recover relevant prior work before making changes and to let work continue across different agents and team members. OrgComputers connects records from systems including GitHub, Linear, Slack, Notion, Jira, Confluence, PagerDuty, and Google Drive through an MCP server. It links entities such as a service to related pull requests, incidents, tickets, conversations, and architecture decisions, while pointing answers back to their sources. The product does not replace those source systems. Handoff can be installed with a shell command that detects Claude Code, Cursor, or Codex. Claude Code sessions are checkpointed in the background; other MCP-compatible clients can ask questions and checkpoint through the hosted endpoint. The install page states that only an agent-written summary is shared, not a raw transcript, and that sharing is limited to people working in the same project. ## OrgComputersでできること ### Shared project workspace Keeps projects, repositories, workstreams, context, tools, activity, members, sessions, documents, code, notes, and links in a workspace available to the team’s agents. ### MCP access for coding agents Works from Claude Code, Cursor, Codex, and other MCP-compatible clients, allowing agents to retrieve and checkpoint information from the workspace. ### Linked entity records Merges different names for the same entity and connects it with related pull requests, tickets, incidents, decisions, conversations, and other records. ### Source-backed retrieval Agents can search, retrieve entities, follow related connections, and trace where a fact came from. Answers point back to the originating sessions or connected sources. ### Historical and permission-aware context Superseded decisions can be marked rather than returned as current, and results respect the viewer’s access—for example, restricted channels are not returned to viewers without permission. ### Agent-written checkpoints The install documentation says that only a short summary written by the agent is shared, rather than the raw session transcript; automatic background checkpointing is specifically identified for Claude Code. ## 利用シーン ### Continue work across coding agents A team can start an investigation in Claude Code, ask about it from Cursor, and finish in Codex while each agent accesses the shared project view. ### Check decisions before changing code Before modifying a service, an agent can look up previous architecture decisions, incidents, pull requests, and follow-up work to avoid reviving an outdated approach. ### Find what the team already tried Developers can ask about prior attempts in a project and receive answers tied to the sessions or records that supplied the information. ### Explore cross-system service history Teams can retrieve an entity such as a service and follow its linked repositories, tickets, incidents, conversations, and decisions across connected systems. ## よくある質問 ### Which coding agents does OrgComputers support? The site names Claude Code, Cursor, and Codex. It also says that any MCP-compatible client can ask questions and checkpoint through the hosted endpoint. ### How is Handoff installed? The install page provides a shell command that detects Claude Code, Cursor, or Codex and configures the client. A teammate joining an existing team uses the same command with the team invite appended. ### Does OrgComputers share raw coding-agent transcripts? The install documentation says that only a short summary written by the agent is shared. It states that raw transcripts do not leave the machine, and that shared information is limited to people working in the same project. ### What happens when information is outdated or restricted? Superseded decisions are marked and are not returned as current. Results also follow permissions, so information from a source the viewer cannot read is not returned. ### Does OrgComputers replace GitHub, Linear, or Slack? No. The product describes those systems as remaining the source of truth; OrgComputers links their records and points answers back to the original sources. ## クイック情報 - **Category**: Developer Tool; AI productivity - **Primary users**: Teams using AI coding agents - **Named clients**: Claude Code, Cursor, and Codex - **Access method**: MCP server plus web app - **Connected systems shown**: GitHub, Linear, Slack, Notion, Jira, Confluence, PagerDuty, and Google Drive - **Pricing**: Not documented on the available product pages