Agent-first · capabilities → public apps

The ecosystem of apps your AI agent can use.

One side is what an agent can do — remember across sessions, recall its memory by meaning, build a knowledge graph, track its tasks, ingest and replay events, stream live changes, and start with zero signup. The other side is the real, public apps and endpoints that grant those abilities: the prime-mcp server and its prime_* tools, the one-click Claude Desktop DXT, the chronis task CLI, the SDKs, and the public event API. Click any capability to see exactly how to wire it.

click a capability to see how to wire it

Agent capability(square)Public app / endpoint(circle)directed edges — hover to read the relation

Loading ecosystem…

The ecosystem your agent can plug into

Purple squares are things your agent can do. Coloured circles are the real, public apps and endpoints that give it those abilities.

Click a capability to see exactly which app(s) provide it — with copy-paste install, MCP config, or curl. Click an app to see its access details and what it unlocks.

Use the filter to focus on MCP servers, CLIs, SDKs, API endpoints, registries, or hosted flows.

Memory over MCP

cargo install allsource-prime gives your agent durable memory — prime_recall, prime_add_node, a knowledge graph, and vector search — local-first, optionally synced to your tenant.

Zero-signup start

An agent can mint a working API key with a single call to /api/v1/agents/anonymous-trial — push events immediately, then claim them into a real tenant via /connect.

Honest install channels

Rust pieces ship on crates.io (allsource-prime, chronis, allsource). The TS / Python / Go SDKs install straight from the GitHub registry — not npm or PyPI.