Your agent

MCP reference

The record as Model Context Protocol tools. Agents read the context of record, propose entries and check authority before acting, under the same labels as every other read.

POST https://YOUR-MCP-DOOR-ADDRESS/mcp

The MCP door is a Model Context Protocol server over streamable HTTP. Any MCP client that supports remote servers can connect.

Authorization

The MCP door follows the MCP authorization specification.

The door publishes protected resource metadata at /.well-known/oauth-protected-resource. Your client reads it, finds the authorization server and signs you in through the browser.
A request without valid authorization gets a 401 with a WWW-Authenticate challenge that points to that metadata, and no tool list.
Your sign-in maps to a Syderial principal and a set of scopes. Roles and groups in your identity provider don’t grant Syderial authority on their own. One person connection reaches the workspaces where that person has membership and grants.
Work without a person present uses an issued credential, such as a hook or service credential. It stays within the issuer’s granted access and can’t ratify as a person. A header credential isn’t a substitute for a person’s browser sign-in.

Sessions

initialize opens a session and returns an Mcp-Session-Id header. Later requests in the session send that header and must authenticate as the same identity. Sessions end on DELETE, when the transport closes, or after a period of inactivity.

Tools

tools/list returns the tools your scopes allow. They fall into these families.

Family	What the tools do
Workspaces	List accessible workspaces and choose a destination for reads and writes.
Record	Read subjects and vocabulary, propose evidence-backed entries and read the history of a subject. Admission posts the lines that pass. A proposal status alone doesn’t prove posting.
Context	Retrieve the context of record for a question as a packet, limited to your clearance.
Evidence	Expand one piece of cited evidence into its signed source window.
Artifacts	Register an artifact, retain source text and read what you’re cleared to see. Retaining text doesn’t propose or post entries.
Case files	Open, update, review and close a bounded working matter, with a receipt on closing.
Authority	Simulate whether an action would be allowed before attempting it.
Briefs	Read the daily and weekly briefs and the items that have drifted, within your boundary.
Execution	Propose an action without side effects, turn an approved proposal into a ticket and read its status.

Syderial checks every tool call against authority and records it. Writes pass through the same stations as every other write, so Syderial commits a proposal only when its evidence supports it.

Execution tools can propose work and record tickets. A ticket doesn’t prove an external action ran. Read its status before reporting an effect.

Keep the record

Use the tool schemas your connection publishes. The agent files the actual source with artifacts_capture_text, reads subjects and meanings with record_context, then writes evidenced lines with claims_propose.

Read per-line outcomes and the receipt. A failed line returns to the agent with a reason. Revise that line or drop it. Don’t repeat posted lines or bypass a failed check. Read back the stored result before saying it posted.

For questions, use search_context. The agent can follow a returned citation with claims_recall or evidence_expand in the same connected session. These reads stay within its access.

claude mcp add --transport http syderial https://YOUR-MCP-DOOR-ADDRESS/mcp

Then run /mcp in a session and sign in.

[mcp_servers.syderial]
url = "https://YOUR-MCP-DOOR-ADDRESS/mcp"

Then run codex mcp login syderial.

See Connect your agent for web and editor clients, and Coding harnesses for gateway configuration.