si.sigit/cli
OfficialResolves repository context from the local Git checkout — it reads the sigit.si Git remote in the current directory to infer the OWNER/NAME so repository-scoped tools default to the repo you're standing in, and defaults a new pull request's head to the checked-out branch.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@si.sigit/clilist the open pull requests"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
sigit-si-cli
si is the command line interface for sigit.si — sign in,
manage repositories, review pull requests, triage issues, search code, and
drive Cloud Sessions, all from the terminal. It's the gh of sigit.si.
Install
cargo install --path crates/cliThis installs a binary named si.
Related MCP server: agent-kernel-mcp
Usage
si auth login # sign in with your sigit.si account
si repo list # your repositories
si repo create my-project --private
# Inside a repo checkout, si infers OWNER/NAME from the sigit.si git remote —
# --repo is only needed when you're not standing in one.
si pr list
si pr view 42
si pr create --title "Fix the thing" --base main
si issue list --state open
si issue create --title "Bug: ..." --body "Steps to reproduce..."
si code search "TODO" --ref main
si code view README.md
si session new --title "Refactor auth"
si billing show
si browse # open the current repo on sigit.si
si api repos # raw escape hatch for any endpointPass --json to any read command for machine-readable output.
As an MCP server
si mcp serves these same commands to an AI agent as
MCP tools over stdio:
{
"mcpServers": {
"sigit": { "command": "si", "args": ["mcp"] }
}
}There is nothing to authenticate — the server reuses the token si auth login
stored. Point your editor at the directory of a checkout and repository tools
default to the repo you're standing in, so the agent can say "list the open
pull requests" without naming one; create_pull_request likewise defaults its
head to the checked-out branch.
Trim the tool list when an agent doesn't need all of it:
si mcp --read-only # drop every tool that writes
si mcp --toolsets repo,issue,pr,code # or SIGIT_TOOLSETS=repo,issueToolsets are account, repo, issue, pr, code, hook, session, and
billing. sigit.si also hosts a remote MCP server at /api/v1/mcp for clients
that would rather connect over HTTP than spawn a binary.
Environments
si --environment dev repo list # a local sigit.si server (localhost:3000)
SIGIT_HOST=staging.example.com si repo listCredentials are stored per-environment in ~/.sigit/token (or
~/.sigit-dev/token). SIGIT_TOKEN overrides the stored token, which is how
CI authenticates without an interactive login.
Workspace layout
crates/sigit-si-api— a standalone, typed Rust client for the sigit.si JSON API. No CLI dependencies (clap, dialoguer, …); usable from any Rust project that wants to talk to sigit.si, including the desktop app.crates/cli— thesibinary: command parsing, terminal rendering, git remote resolution, and credential storage, built on top ofsigit-si-api.
Issues, pull requests, and code search have no REST surface on sigit.si yet —
they're served over the same MCP (Model Context Protocol) JSON-RPC endpoint
the AI agent uses. sigit-si-api speaks that transport internally; it's not
visible at the call site.
MCP Registry listings
This repo publishes two entries to the official MCP Registry, one per way of reaching the same tools, so registry-aware clients can add either in one click:
Listing | File | What it is |
|
| This CLI: a cargo package listing that runs |
|
| The hosted server at |
The CLI holds server.json because it's what this repo ships, and that's the
filename mcp-publisher defaults to. Sibling listings are named after their
own leaf, so the file says which entry it publishes.
Both go out through the release-mcp-registry.yml workflow, and both prove
namespace ownership with a DNS TXT record on sigit.si rather than GitHub
OIDC, so one credential covers the pair. A package listing has a second hurdle:
the registry fetches the crate from crates.io and looks for the
mcp-name: si.sigit/cli marker in its README, which is why that marker sits in
this file. Crate versions are immutable, so a release that ships without the
marker can never be listed, and fixing it costs a new version. The internal
setup and release runbook live with the server (private sigit-si repo).
Development
make build # cargo build --workspace
make test # cargo test --workspace
make lint # cargo clippy --workspace --all-targets -- -D warnings
make fmt # cargo fmt --all
make check # fmt + lint + testDisclaimer
All em dashes are written by Seto Elkahfi.
License
Apache-2.0
This server cannot be deployed
Maintenance
Related MCP Connectors
Persistent memory, hybrid search and a goal graph for AI agents, over stdio or remote HTTP.
Develop, manage, and debug Railway projects, services, and deployments from within agents.
Git-backed platform for skills, tools, and context for AI agents
Read a project's prompts, logs and agents, and send new work to the agent on your own machines.
Related MCP Servers
- AlicenseAqualityBmaintenanceEnables an AI agent to run local tools on the user's own machine via stdio, including command execution, workspace file read/write, and system status checks.52MIT
- AlicenseBqualityBmaintenanceEnables IDE and DeepSeek Harness agents to control an agent-kernel instance over stdio, exposing tools for auth, projects, assignments, runs, scheduler nudges, and executor settings.12MIT
- FlicenseNot gradedqualityCmaintenanceServes local-first project memory over stdio, exposing tools to inspect project state, search and retrieve context, report memory, and create Git-aware checkpoints for coding agents.1-
- AlicenseNot gradedqualityAmaintenanceServes local coding-agent session history over stdio, letting MCP-capable agents search past sessions across Codex, Claude Code, Gemini CLI, OpenCode, jcode, and Pi and read bounded, secret-redacted transcripts. Read-only mode restricts it to search and transcript tools, while write-enabled mode can create copies of a conversation in another agent's native format.51MIT