| What it is | A company brain for a team's coding agents. It learns the rules your team decides and serves the ones that apply to each session. | A context database for AI agents. It keeps memories, documents and skills as files under viking:// paths, plus plugins that add memory to coding agents. |
|---|
| Where the knowledge comes from | Slack, GitHub, GitLab, Notion, Confluence, Jira, Linear, Google Drive and Granola, plus what agent sessions record. | What happens in agent sessions: its coding plugins capture every turn, tool output included, and extract memories as Markdown when the session is committed. You can also add documents. |
|---|
| Who decides what gets in | You set the gate. Manual holds everything for a person, Hybrid adds routine rules and holds the important ones, Auto adds everything else. A contradiction always waits for a person. | Extraction is automatic, and the Claude Code plugin documents no review step. Memories are private: to share one, its owner publishes a copy, and access rules decide who can read it. |
|---|
| How it reaches the agent | Hooks add the rules that apply to the session before the agent writes a line. The same rules are on an MCP server, the CLI and /harbor in Slack. | In Claude Code, hooks add your profile and memory index when a session starts and search for relevant memories before every prompt. Agents can also use its MCP server, the ov CLI, SDKs or HTTP. |
|---|
| When a decision changes | When a push deletes the file a rule is about, the rule is retired, or flagged if other repos still use it. A newer decision that contradicts a rule waits for a person. | Each repo keeps one memory across clones and worktrees. Duplicates merge when you run its consolidation command, which updates one memory and deletes the others. |
|---|
| Seeing what was used | Every rule shows how often it was served and how often an answer cited it. | In Claude Code, a card under each answer lists the OpenViking sources it drew on, and a status command shows recall counts. |
|---|
| Stopping a risky command | Block rules refuse a command before it runs, in Claude Code, Cursor, Codex and GitHub Copilot, and hand the agent something to run instead. | Not part of the product. |
|---|
| Agents | Claude Code, Cursor, Codex and GitHub Copilot (the CLI and VS Code). | Claude Code, Codex, Cursor and TRAE through hooks and MCP, plus OpenCode, OpenClaw, Hermes and any MCP client. |
|---|
| Open source and hosting | The CLI, harborloop, is MIT licensed on npm. The service is hosted. | Open source under AGPLv3 and self-hosted. Volcengine also runs it as a hosted service, and sells a self-managed commercial edition. |
|---|
| Price | Setup is self-serve and takes two minutes, with no call required. Free costs nothing for your whole team on one source, Pro is $24 per person a month billed yearly or $30 monthly, and Premium is $36, or $45 monthly. | Self-hosted is free. Volcengine's hosted service is free for the first 50 files, then has Personal and Enterprise plans. |
|---|
| Best for | Engineering teams who want every coding session to start with what the team decided. | Developers who want a self-hosted context store their agents fill and search on their own. |
|---|