Pi-Bolt
Pi, the coding agent, compiled ahead of time to native code. Starts 3× sooner than Pi on Bun and uses about a third of its CPU.
Install / Use
npx skills add opensec-git/Pi-BoltInstalls into whichever agent you are using.
Other
Other agent config
Quality Score
Category
AI & Machine LearningSupported Platforms
Tags
Our assessment of Pi-Bolt
Pi-Bolt scores 91/100 on our quality scale, 285th of 965 AI & Machine Learning skills we index (top 30%).
Its Other is 16 KB long, well organised into 12 sections with 7 code examples: a thorough specification that gives an agent plenty to work with.
It has 315 GitHub stars, a meaningful sign that others use it.
Maintenance, license and trust
- The repository was last updated today, so Pi-Bolt is actively maintained.
- It is released under the MIT license, a permissive license that allows use, modification and commercial use with attribution.
- Its trust signals score 100/100, with no cautions. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.
Safety scan
No issues foundOur scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands (3 minor notes below).
- noteInstalls by piping a downloaded script into a shellline 43
curl -fsSL https://pi-bolt.opensec.in/install.sh | sh - noteInstalls by piping a downloaded script into a shellline 49
irm https://pi-bolt.opensec.in/install.ps1 | iex - noteInstalls by piping a downloaded script into a shellline 55
powershell -c "irm https://pi-bolt.opensec.in/install.ps1 | iex"
Automated pattern scan on 2026-10-13. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.
Pi-Bolt compared with similar skills
All 4 of these similar skills score higher than Pi-Bolt; compare them before choosing.
| Skill | Score | Stars | Updated | Format |
|---|---|---|---|---|
| Pi-Bolt (this skill)by opensec-git | 91 | 315 | today | Other |
| claude-memby thedotmack | 100 | 100.3k | today | CLAUDE.md |
| Understand-Anythingby Egonex-AI | 100 | 86.1k | 1d ago | CLAUDE.md |
| headroomby headroomlabs-ai | 100 | 75.2k | today | CLAUDE.md |
| CowAgentby zhayujie | 100 | 47.4k | today | CLAUDE.md |
Frequently asked questions
- How do I install Pi-Bolt?
- Run
npx skills add opensec-git/Pi-Bolt. The install tabs above show the steps for each supported agent. - Which AI agents does Pi-Bolt work with?
- It is written for Universal, as a Other file. Other agents that read the same format can often use it too.
- Is Pi-Bolt safe to use?
- Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands (3 minor notes below). It is MIT-licensed and scores 100/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
- Is Pi-Bolt still maintained?
- The repository was last updated today, so Pi-Bolt is actively maintained.
Skill content
View source on GitHubPi-Bolt is the Pi you already use, with the same commands, sessions, settings, providers and extensions. Every function is compiled to machine code when the executable is built, so nothing is parsed or JIT-compiled at launch. It starts two to three times sooner than Pi on Bun, uses a quarter to a third of its CPU over a session, and handles long answers and large files six to seventy-five times more cheaply.
<picture> <source media="(prefers-color-scheme: dark)" srcset="docs/images/bench-hero-dark.svg"> <img alt="Pi-Bolt 0.8.5 vs Pi on Bun 1.4.2 vs Pi on Node 24: ready to type 38 / 116 / 296 ms; CPU per session 193 / 746 / 1,093 ms; CPU while streaming 204 / 382 / 475 ms; memory after a long session 134 / 253 / 536 MB" src="docs/images/bench-hero-light.svg"> </picture><sub>AMD EPYC 7B13, Linux. Pi-Bolt 0.8.5 against Pi 1.1.0 as released on stock Bun 1.4.2 and Node 24; medians of interleaved runs, lower is better. See Benchmarks.</sub>
[!NOTE] Pi-Bolt is an independent fork of Pi, not affiliated with Pi's authors or with Oven.
Install
curl -fsSL https://pi-bolt.opensec.in/install.sh | sh
On Windows, in PowerShell:
irm https://pi-bolt.opensec.in/install.ps1 | iex
or in Command Prompt:
powershell -c "irm https://pi-bolt.opensec.in/install.ps1 | iex"
or npm install -g --allow-scripts=pi-bolt pi-bolt, or bun add -g --trust pi-bolt. Then run pi-bolt in your project. Your
Pi configuration in ~/.pi/agent is used as is; new to Pi? Run /login to connect a provider. pi-bolt update installs the
latest release.
The installer verifies the release signature and offers three optional extensions by OpenSec:
| Extension | What it does |
|---|---|
| opensec-pi-subagents | Specialized agents in separate sessions, parallel workflows and scheduled jobs |
| opensec-pi-todo | A todo list for the model, shown as a live panel above the editor |
| opensec-pi-fff | pi-fff's fast fuzzy find, grep and @-mentions, at full speed on the default build |
JIT build, for heavy plugins
The default build has no JIT: Pi's own code is compiled machine code, and an extension loaded at run time is interpreted. Light
extensions don't notice. For extensions that do heavy work (search, parsing) or use bun:ffi, install the JIT
build. Pi itself stays compiled, so it is within about 5% of the default.
curl -fsSL https://pi-bolt.opensec.in/install.sh | PIBOLT_VARIANT=arm64-jit sh # macOS
curl -fsSL https://pi-bolt.opensec.in/install.sh | PIBOLT_VARIANT=x64-jit sh # Linux
On Windows, run $env:PIBOLT_VARIANT = "x64-jit" before the PowerShell command. With npm, keep PIBOLT_VARIANT set in your
shell profile; it is read on every run. pi-bolt --version then ends in JIT on, and pi-bolt update keeps the build you chose.
Download a build from the latest release, check it against
SHA256SUMS, unpack it and run ./pi (on Windows, pi-bolt.exe).
| Download | For |
|---|---|
| pi-bolt-linux-x64.tar.xz | Linux, CPUs with AVX2 (Intel Haswell, AMD Zen and later) |
| pi-bolt-linux-x64-baseline.tar.xz | Linux, any x86-64 CPU |
| pi-bolt-darwin-arm64.tar.xz | macOS, Apple silicon |
| pi-bolt-win32-x64.zip | Windows x64; compiled code on CPUs with AVX2, bytecode on others |
-jit builds are for heavy use of plugins loaded at run time; the runtime archive is for compiling plugins in.
Requirements
- Linux x86-64 with glibc 2.17 or later (Ubuntu 20.04+, Debian 11+, RHEL/Rocky 8+, Amazon Linux 2). musl is not supported.
- macOS 13 or later on Apple silicon. Install with the installer or npm (Troubleshooting).
- Windows 10 (1809) or later, or Windows 11, on x64. Install with the installer or npm; for problems, see Troubleshooting: Windows.
- Linux and Windows on ARM64 are not available yet.
Pi-Bolt installs next to Pi as pi-bolt, so both can stay installed. Running the installer again updates, reinstalls or
uninstalls it.
Benchmarks
Pi-Bolt 0.8.5 against Pi 1.1.0 as released, on stock Bun 1.4.2 and on Node, against two other fast Pi builds, PiG 0.4.1, the Go port of Pi (based on Pi 1.0.3), and oh-my-pi 18.8.7, a fork of Pi on Bun, and against OpenCode 2.0. Runs are interleaved on the same machine; medians, lower is better.
Linux x86-64 (AMD EPYC 7B13)
| Benchmark | Pi-Bolt | PiG | oh-my-pi | OpenCode 2.0 | Pi on Bun | Pi on Node 24 |
|---|---:|---:|---:|---:|---:|---:|
| Writing a 200 KB file through a tool call | 0.6 s | 77.5 s | 3.3 s | 2.4 s | 24.0 s | 32.5 s |
| CPU, streaming a 60,000-character answer | 2.6 s | 16.7 s | 2.2 s | 8.6 s | 40.2 s | 38.7 s |
| Memory after a 4.2M-token session | 134 MB | 349 MB | – | 273 MB¹ | 253 MB | 536 MB |
| Memory of a session in tmux | 22 MB | 48 MB | 334 MB | –² | 84 MB | 185 MB |
| CPU, interactive session (5 prompts) | 193 ms | 2,458 ms | 2,800 ms | 8,593 ms | 746 ms | 1,093 ms |
| CPU, one prompt (pi -p) | 61 ms | 227 ms | 1,524 ms | 2,220 ms | 283 ms | 550 ms |
| Peak memory, one prompt (pi -p) | 123 MB | 55 MB | 505 MB | 346 MB | 147 MB | 135 MB |
| Ready to type | 38 ms | 31 ms | 890 ms | 1,687 ms | 116 ms | 296 ms |
macOS (Apple M5)
| Benchmark | Pi-Bolt | PiG | oh-my-pi | OpenCode 2.0 | Pi on Bun | Pi on Node 26 |
|---|---:|---:|---:|---:|---:|---:|
| Writing a 200 KB file through a tool call | 0.4 s | 56.8 s | 2.5 s | 1.7 s | 22.6 s | 28.1 s |
| CPU, streaming a 60,000-character answer | 6.5 s | 16.5 s | 5.1 s | 13.6 s | 37.9 s | 37.6 s |
| Memory after a 4.2M-token session | 38 MB | 330 MB | – | 260 MB¹ | 92 MB | 1,581 MB |
| Memory of a session in tmux | 24 MB | 55 MB | 330 MB | –² | 65 MB | 162 MB |
| CPU, interactive session (5 prompts) | 139 ms | 2,367 ms | 2,374 ms | 6,493 ms | 561 ms | 937 ms |
| CPU, one prompt (pi -p) | 49 ms | 230 ms | 1,030 ms | 1,548 ms | 221 ms | 512 ms |
| Peak memory, one prompt (pi -p) | 70 MB | 50 MB | 419 MB | 379 MB | 120 MB | 138 MB |
| Ready to type | 41 ms | 39 ms | 612 ms | 1,489 ms | 100 ms | 348 ms |
PiG starts a little sooner and its single-prompt process is smaller; oh-my-pi streams a long answer with a little less CPU.
Pi-Bolt uses the least CPU in sessions and per prompt, the least memory over a session, and writes large files fastest.
oh-my-pi compacts a long conversation on its own, so the long session does not compare it (–). OpenCode 2.0.27 (@opencode/cli)
runs with its own server in the same process tree (--standalone), in a separate run next to Pi-Bolt on the same machines.
¹ The process started, without OpenCode's server process: its memory is higher. ² OpenCode did not finish its answers in tmux.
Windows x64 (Intel Core i5-1335U laptop; Pi-Bolt 0.8.5, Pi 1.1.0 as released)
| Benchmark | Pi-Bolt | Pi on Bun 1.4.2 | Pi on Node 24 |
|---|---:|---:|---:|
| Writing a 200 KB file through a tool call | 0.5 s | 38.1 s | 26.6 s |
| CPU, streaming a 60,000-character answer | 6.9 s | 41.1 s | 32.6 s |
| Memory after a 4.2M-token session | 122 MB | 248 MB | 413 MB |
| Memory of a session in a terminal (ConPTY) | 32 MB | 97 MB | 60 MB |
| CPU, interactive session (5 prompts) | 278 ms | 805 ms | 1,125 ms |
| CPU, one prompt (pi -p) | 92 ms | 329 ms | 544 ms |
| Ready to type | 105 ms | 177 ms | 311 ms |
Method, raw data and the macOS and Windows charts: [docs/BENCHMARKS.md](docs/BENCHMARKS
Truncated for display — read the full file on GitHub.
Related Skills
claude-mem
100.3kPersistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More
Understand-Anything
86.1kGraphs that teach > graphs that impress. Turn any code into an interactive knowledge graph you can explore, search, and ask questions about. Works with Claude Code, Codex, Cursor, Copilot, Gemini CLI, and more.
headroom
75.2kCompress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.
CowAgent
47.4kOpen-source personal AI assistant & Agent Harness. Plans tasks, runs tools and skills, self-evolves with memory and knowledge. Multi-agent, multi-model, multi-channel. Lightweight, extensible, one-line install.
Languages
Trust signals
From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.