npm package discovery and stats viewer.

Discover Tips

  • General search

    [free text search, go nuts!]

  • Package details

    pkg:[package-name]

  • User packages

    @[username]

Sponsor

Optimize Toolset

I’ve always been into building performant and accessible sites, but lately I’ve been taking it extremely seriously. So much so that I’ve been building a tool to help me optimize and monitor the sites that I build to make sure that I’m making an attempt to offer the best experience to those who visit them. If you’re into performant, accessible and SEO friendly sites, you might like it too! You can check it out at Optimize Toolset.

About

Hi, 👋, I’m Ryan Hefner  and I built this site for me, and you! The goal of this site was to provide an easy way for me to check the stats on my npm packages, both for prioritizing issues and updates, and to give me a little kick in the pants to keep up on stuff.

As I was building it, I realized that I was actually using the tool to build the tool, and figured I might as well put this out there and hopefully others will find it to be a fast and useful way to search and browse npm packages as I have.

If you’re interested in other things I’m working on, follow me on Twitter or check out the open source projects I’ve been publishing on GitHub.

I am also working on a Twitter bot for this site to tweet the most popular, newest, random packages from npm. Please follow that account now and it will start sending out packages soon–ish.

Open Software & Tools

This site wouldn’t be possible without the immense generosity and tireless efforts from the people who make contributions to the world and share their work via open source initiatives. Thank you 🙏

© 2026 – Pkg Stats / Ryan Hefner

@taf0711/splice

v0.2.0

Published

Splice — a fast, multi-provider terminal coding agent.

Readme

Splice is a local-first coding agent for people who want more than a model in a shell loop. It can inspect a repository, edit files, run checks, and use the model and provider you choose. For substantial work, it classifies the request, assembles only the context a stage needs, and routes the change through a specialized execution plan.

Start in three commands

npm install -g @taf0711/splice
splice

Then describe the change. For a direct, scriptable run:

splice exec "fix the failing test in ./pkg"

The first launch opens setup for a provider and model. See Install for source builds, platform notes, and the optional memory sidecar.

One-line install

Use the Bash installer on Linux or macOS:

curl -fsSL https://raw.githubusercontent.com/Taf0711/splice/main/scripts/install.sh | bash
wget -qO- https://raw.githubusercontent.com/Taf0711/splice/main/scripts/install.sh | bash

Use the plain PowerShell form on Windows:

irm https://raw.githubusercontent.com/Taf0711/splice/main/scripts/install.ps1 | iex

Use the PowerShell argument form to set a version:

& ([scriptblock]::Create((irm https://raw.githubusercontent.com/Taf0711/splice/main/scripts/install.ps1))) -Version 0.2.0

Why Splice exists

Most coding agents have one long conversation responsible for understanding, editing, testing, and deciding whether to continue. Splice separates those jobs without turning them into a noisy agent swarm.

  • A typed pipeline. Stages exchange validated Go structures, not loose chat transcripts. The orchestrator decides what each stage receives and what runs next.
  • Deterministic first. Repository reads, search, AST checks, static analysis, tests, diffs, and exit codes come from local tools. The model is used where judgment is needed, not where a command is more reliable.
  • Trajectory awareness. The run monitor detects hard limits, repeated state, oscillation, regression, and collapsing confidence. A stuck run surfaces the problem instead of spending tokens forever.
  • Least context by default. The original request is distilled before it reaches later stages. Each stage receives a small summary and the evidence it actually needs.
  • Zero's safety substrate. File writes, shell commands, network access, and additional write roots still pass through Zero's permissions, sandbox, hooks, and tool registry.
  • Local by default. Sessions and optional memory stay on disk. Splice does not add telemetry or require a hosted coordinator.

The Splice loop

request
   │
   ▼
classify ──► typed execution plan ──► focused context
                                      │
                                      ▼
             write ──► analyze ──► test ──► audit
                                      │
                                      ▼
                         trajectory decision
                         continue, escalate, or stop

The pipeline has five request tiers, from trivial to architectural. A normal run may use a model-backed code writer and test generator, while static analysis, security checks, and test execution stay deterministic and local. Every stage has a contract and validation. Malformed output is an error with its offending field, not an empty fallback.

Choose your mode

Interactive

splice

Use the TUI when you want a live conversation, provider and model pickers, permission prompts, plan review, themes, image input, session resume, or forked exploration.

A new session starts in design mode. Describe the work, use /crystallize to turn the conversation into a typed plan, then /approve to execute it. If the plan is already in your head, use /exec <prompt> and go straight to the pipeline.

These controls set the appearance:

| Control | Effect | |---|---| | NO_COLOR=<anything> | disables color output | | SPLICE_THEME=<name> | selects the startup theme (auto, dark, light, or a color theme like dracula, nord, gruvbox, tokyo-night, catppuccin, one-dark, solarized-dark, rose-pine, everforest, solarized-light) | | --theme <name> | selects the TUI theme from the CLI (same names) | | /theme | opens the theme picker inside the TUI (live preview; /theme <name> switches directly) | | SPLICE_NO_FADE=1 | disables streaming fade animation |

Meaning does not rely on color alone. Diffs, permissions, and statuses also use text or glyph markers.

Headless

splice exec "explain internal/agent/loop.go"
splice exec --use-spec "add rate limiting to the API client"
splice exec --worktree "try the migration in isolation"
splice exec --worktree --merge-back "run isolated, then merge the result back"
splice exec --plan design-plan.json

splice exec supports text, JSON, and stream-JSON input and output, isolated git worktrees, resumable sessions, and meaningful exit codes for automation. For a streaming client:

splice exec \
  --input-format stream-json \
  --output-format stream-json < turns.jsonl

The event contract lives in docs/STREAM_JSON_PROTOCOL.md.

Providers and local models

Bring the provider that fits the job. Splice supports OpenAI, Anthropic, Gemini, Groq, OpenRouter, DeepSeek, Mistral, xAI, Qwen, Kimi, GitHub Models, Ollama, LM Studio, and OpenAI- or Anthropic-compatible endpoints.

splice setup
splice providers list
splice models list
splice doctor

Set a provider key before setup or enter it in the wizard:

export OPENAI_API_KEY=sk-...
export ANTHROPIC_API_KEY=...
export GEMINI_API_KEY=...

Local models work through Ollama or LM Studio. Model-backed stages require tool calling. If a provider returns an invalid typed payload, Splice retries the correction and then reports an actionable error. It does not silently switch a local run to a cloud provider.

Safety is part of the workflow

Splice makes side effects visible instead of hiding them behind an autonomous mode:

  • workspace reads are allowed by default;
  • writes stay inside the workspace unless you grant another root;
  • shell commands, network access, destructive actions, and elevated actions are permission-gated;
  • --add-dir <path> grants a specific additional write root;
  • unsafe and autonomous modes are explicit opt-ins;
  • sandbox policy and grants can be inspected before a run.
splice sandbox policy
splice sandbox grants list
splice exec --add-dir ../shared "update both repositories"

Read SECURITY.md for the threat model and reporting process.

A small CLI surface for big workflows

| Command | What it is for | |---|---| | splice | Interactive TUI | | splice exec | One-shot and scripted runs | | splice setup | First-run provider setup | | splice providers / models | Provider profiles and model capabilities | | splice doctor | Setup and connectivity checks | | splice spec | Spec-mode drafts | | splice sessions | Resume, fork, and inspect sessions | | splice specialist | Specialist subagents | | splice skills / plugins / hooks | Extend the agent locally | | splice mcp | Configure MCP servers and tools | | splice worktrees | Prepare isolated git worktrees | | splice verify | Find and run repository checks | | splice update / upgrade | Keep the binary current |

Documentation

For contributors

Splice-specific pipeline code lives in internal/splice/. It is pure Go and keeps provider SDKs out of the pipeline layer. The optional memory sidecar is a separate module in memd/ and communicates over a Unix socket.

go test ./...
go vet ./...
gofmt -l .
cd memd && go test ./...

Read CONTRIBUTING.md before opening a change. The project is currently using an issue-first contribution policy.

Built on Zero's Engine

Splice is built on top of Gitlawb's Zero Engine, an open-source MIT-licensed terminal coding agent. Zero provides the TUI, provider adapters, sessions, tools, sandbox, permissions, worktrees, MCP, skills, plugins, and hooks. Splice keeps that foundation and adds the orchestrator-mediated pipeline, typed stage contracts, deterministic checks, trajectory monitoring, and the optional memory sidecar.

Upstream Zero code retains its original copyright and license. See LICENSE and SECURITY.md.

License

Splice is released under the MIT License.