Install Causal in your agent

Causal installs into eight agent runtimes:

Claude CodeCursorCodexOpenClawHermesGitHub CopilotGemini CLIOpenCode

All eight take the memory skills. The first five also get MCP wiring and the always-on rule today; Copilot, Gemini CLI, and OpenCode get those as each one’s config support settles.

Run it

From the project you want your agent to remember:

npx causaldotstore

That signs you in, finds the agent runtimes you already have, and wires Causal into each one. You keep the agent you work in.

What you get back

$ npx causaldotstoreInstalled 2 skill(s) into 3 runtime(s):  Claude Code      → ~/.claude/skills  (2 skills)                   + 4 commands → ~/.claude/commands                   MCP → ~/.claude.json  (wired)                   rule → ~/.claude/CLAUDE.md  (appended)  Codex CLI        → ~/.agents/skills  (2 skills)                   MCP → ~/.codex/config.toml  (wired)                   rule → ~/.codex/AGENTS.md  (appended)  Cursor           → ~/.cursor/skills  (2 skills)                   MCP → ~/.cursor/mcp.json  (wired)                   Cursor rules are project-level — rerun with --project inside a repoSkills: causal-memory-capture, causal-memory-recall

Read the summary as three claims per runtime. Skills are the memory skills themselves, copied into the runtime's skills directory. MCP is the connection: how the agent reaches the experience graph to read precedent and write outcomes. Rule is the trigger, appended to the file that runtime already loads every session.

The rule is the one that matters at runtime. Without it, recall is something the agent has to remember to ask for, and it arrives after the agent has already chosen a direction. With it, the relevant precedent is in context when the agent starts deciding, and a prior attempt comes back with the outcome it caused. The longer account is on Recall and capture.

Rerunning is safe. Every write is idempotent: the rule carries a marker, so a second run reports already present rather than appending twice.

Pick your runtimes

npx causaldotstore installs into everything it detects. Name them instead to skip the wizard:

npx causaldotstore --runtimes=claude,cursor,codex

Add --project to install into the current repo rather than your home directory, and --dry-run to print the plan without writing anything.

Claude Code additionally gets four slash commands.

OpenClaw and Hermes are wired through their own CLIs — openclaw mcp set and hermes mcp add — because OpenClaw keeps its config in JSON5 and Hermes keeps MCP inside its shared settings YAML. Rewriting either by hand would drop the comments and structure in a file you own. If the runtime is not on PATH when you install, Causal prints the one command to finish it.

What it costs to start

A personal workspace starts with 2,000 memories, no card and no deadline. Team workspaces start on a 14-day card-less Pro trial with 10,000 memories. Reads are never billed. The full meter is on the pricing page.

Check it landed

npx causaldotstore status

That prints the key on file, the workspace link, and which runtimes are wired.

Handoff between runtimes

They read and write the same graph, so what one agent recorded is what the next one reads. A long Claude Code session that ends at a usage limit leaves its experience behind, and a fresh Cursor session picks the work up from that experience rather than from a re-paste of the transcript.

The second agent inherits prior actions and their outcomes, not a copy of the earlier context. A precedent arrives with the outcome it caused, so it can strike a plan today instead of sitting in a search index.

Follow-up work stays attached to what caused it, so a change made in Codex after a decision taken in Claude Code traces back to that decision instead of starting a new history. AI agent governance and audit trail covers who may read which entries, how long they are kept, and how a served precedent traces back to the work that produced it.

What if my agent is not on that list?

Python and TypeScript SDKs cover anything outside the eight.

npm install causal

pip install causal

Both are under the same pre-launch caveat as the installer: the package names are settled, but nothing is published yet.

Not yet published

The SDK surface — client construction, method names, and example code — lands here at launch. So does the rule's own text: what it says, and how to edit or verify it by hand.