ai-coder 1.5.0 — AI Coding & DevOps Agent
.aic095{–c:#22d3ee;–b:#3b82f6;–g:#4ade80;–line:#293443;–card:#121923;–muted:#aab5c3;color:#e9eef5;font-family:Inter,-apple-system,BlinkMacSystemFont,”Segoe UI”,sans-serif;line-height:1.65}.aic095 *{box-sizing:border-box}.aic095 a{text-decoration:none}.aic-shell{max-width:1080px;margin:0 auto}.aic-hero{padding:64px 40px 40px;border:1px solid var(–line);border-radius:24px;background:radial-gradient(circle at 88% 12%,rgba(34,211,238,.18),transparent 30%),linear-gradient(145deg,#0c1118,#121a25)}.aic-badge{display:inline-flex;gap:8px;align-items:center;padding:7px 11px;border:1px solid rgba(74,222,128,.38);border-radius:999px;background:rgba(74,222,128,.08);color:#86efac;font-size:.78rem;font-weight:800;letter-spacing:.08em;text-transform:uppercase}.aic-dot{width:8px;height:8px;border-radius:50%;background:var(–g);box-shadow:0 0 14px var(–g)}.aic-hero h2{margin:18px 0 12px;color:#fff;font-size:clamp(2.6rem,6vw,5rem);line-height:.98;letter-spacing:-.05em}.aic-hero h2 span{background:linear-gradient(90deg,var(–c),var(–b));-webkit-background-clip:text;background-clip:text;color:transparent}.aic-lead{max-width:800px;margin:0;color:#c7d1dd;font-size:1.12rem}.aic-actions{display:flex;flex-wrap:wrap;gap:11px;margin-top:25px}.aic-btn{display:inline-flex;padding:13px 17px;border-radius:10px;font-weight:800}.aic-primary{background:linear-gradient(110deg,var(–b),#2563eb);color:#fff!important}.aic-secondary{border:1px solid #38506a;color:#e8eef6!important}.aic-facts{display:grid;grid-template-columns:repeat(4,1fr);gap:1px;margin-top:16px;overflow:hidden;border:1px solid var(–line);border-radius:14px;background:var(–line)}.aic-fact{padding:16px;background:#101720}.aic-fact small{display:block;color:#7f8b9a;font-size:.7rem;font-weight:800;text-transform:uppercase;letter-spacing:.08em}.aic-fact strong{display:block;margin-top:4px;color:#f6f8fb}.aic-section{padding:56px 4px 0}.aic-eye{color:var(–c);font-size:.77rem;font-weight:800;letter-spacing:.12em;text-transform:uppercase}.aic-section h3{margin:7px 0 10px;color:#f7f9fc;font-size:clamp(1.75rem,4vw,2.5rem);line-height:1.13}.aic-section-lead{max-width:820px;margin:0 0 23px;color:var(–muted)}.aic-grid{display:grid;grid-template-columns:repeat(2,1fr);gap:13px}.aic-card{padding:22px;border:1px solid var(–line);border-radius:15px;background:linear-gradient(145deg,var(–card),#0f151e)}.aic-card h4{margin:0 0 7px;color:#f2f6fa}.aic-card p{margin:0;color:var(–muted);font-size:.94rem}.aic-download{display:grid;grid-template-columns:minmax(0,1.3fr) minmax(280px,.7fr);gap:15px;padding:23px;border:1px solid rgba(59,130,246,.34);border-radius:17px;background:rgba(59,130,246,.055)}.aic-file,.aic-code,.aic-check code{font-family:”JetBrains Mono”,ui-monospace,monospace}.aic-file{padding:11px 13px;border:1px solid var(–line);border-radius:9px;background:#090e14;color:#dce7f2;overflow-wrap:anywhere}.aic-code{margin:11px 0 0;padding:14px;border:1px solid var(–line);border-radius:10px;background:#080c11;color:#b8f3dc;font-size:.82rem;line-height:1.65;overflow:auto;white-space:pre-wrap}.aic-check code{display:block;margin-bottom:9px;padding:11px;border:1px solid var(–line);border-radius:9px;background:#080c11;color:#91e6c2;font-size:.74rem;overflow-wrap:anywhere;white-space:normal}.aic-note{color:#8f9baa;font-size:.84rem}.aic-warning{margin-top:14px;padding:14px 16px;border-left:3px solid #fbbf24;border-radius:7px;background:rgba(251,191,36,.07);color:#decda7;font-size:.88rem}.aic-flow{display:grid;grid-template-columns:repeat(5,1fr);gap:9px}.aic-step{padding:16px 13px;border:1px solid var(–line);border-radius:13px;background:var(–card);text-align:center}.aic-step strong{display:block;color:#fff}.aic-step span{color:var(–muted);font-size:.82rem}.aic-faq{display:grid;gap:9px}.aic-faq details{padding:0 16px;border:1px solid var(–line);border-radius:11px;background:var(–card)}.aic-faq summary{cursor:pointer;padding:15px 0;color:#edf3f8;font-weight:750}.aic-faq p{margin:0;padding:0 0 16px;color:var(–muted)}.aic-final{margin:56px 0 8px;padding:34px;border:1px solid rgba(34,211,238,.28);border-radius:19px;background:#10161e;text-align:center}.aic-final h3{margin:0 0 8px;color:#fff;font-size:1.95rem}.aic-final p{max-width:680px;margin:0 auto 18px;color:var(–muted)}
@media(max-width:820px){.aic-hero{padding:48px 24px 30px}.aic-facts{grid-template-columns:repeat(2,1fr)}.aic-grid,.aic-download{grid-template-columns:1fr}.aic-flow{grid-template-columns:1fr 1fr}}@media(max-width:480px){.aic-facts,.aic-flow{grid-template-columns:1fr}.aic-actions .aic-btn{width:100%}}
ai-coder 1.5.0
A coding and DevOps agent for AILinux and TriForce. ai-coder combines a local CLI and PyQt6 desktop interface with MCP tools, provider-independent model routing, secure Bring Your Own Key (BYOK), controlled system access, and an autonomous agent loop that can inspect, change, test, and verify real systems.
A real system agent, not just a code completion box.
The local client stays thin while TriForce provides authentication, model routing, provider abstraction, and the MCP control plane. The agent can work directly with repositories, files, services, logs, containers, and development tooling on the machine — subject to the configured approval policy.
Autonomous tool loop
Inspect → plan → execute → verify. Multi-step coding and DevOps work can continue until the task is actually checked.
MCP system tools
Connect models to real tools for code, files, Git, diagnostics, services, containers, search, remote systems, and more.
Provider-independent routing
Primary and fallback models stay separate, so a failed provider can be replaced by a real alternative route.
Local control stays local
Writes and elevated operations pass through the local approval and privilege broker; sudo passwords stay inside the OS.
Project Memory: current project state that survives sessions.
Structured Project Memory
Store todos, decisions, architecture, bugs, lessons, features, summaries and documentation as current project state instead of reconstructing them from old chats.
Local-first & offline
Project Memory lives in the existing evidence database. Offline changes stay dirty locally and synchronize automatically when TriForce is reachable again.
Conflict-safe sync
Optimistic concurrency, server sequence numbers and tombstones preserve conflicting revisions instead of silently overwriting another machine.
Claude-Mem history
Accepted revisions are mirrored into Claude-Mem as episodic history while Project Memory remains the authoritative current state.
Reinstall recovery
Stable project identity lets a fresh AICoder installation log in, recognize the project and restore synchronized project state from TriForce.
Nova Voice
Version 1.5.0 adds the tested Nova Voice path for local speech input while retaining Project Memory, BYOK, CLI and GUI workflows.
Your provider. Your API key. Your model.
ai-coder 1.5.0 can route model calls directly through API keys stored securely in the operating-system keyring. Keys are not written to state.json, chat history or logs. Without a configured direct key, the existing TriForce route remains available.
Direct providers
OpenAI, Anthropic, Google / Gemini, OpenRouter, Mistral, Groq, Cerebras and NVIDIA are supported directly.
Every main runtime
BYOK routing now applies to ask, chat, task, review, autonomous agent, PyQt6 GUI and multi-agent team runs.
OS-keyring secrets
Provider credentials stay in KWallet / Secret Service or the platform keyring instead of AICoder project state.
CLI + GUI
Manage keys in the GUI or with aicoder credentials set <provider> and remove them with aicoder credentials delete <provider>.
ai-coder 1.5.0 for Linux & Windows
The production 1.5.0 Linux and Windows packages are hosted directly on ailinux.me. The Linux .deb is also the current APT candidate in the AILinux repository.
Debian package
81.8 MiB · AMD64 · published September 29, 2026
wget https://ailinux.me/downloads/AI-Coder/aicoder_1.5.0_amd64.deb
sudo apt install ./aicoder_1.5.0_amd64.deb
aicoderAILinux:sudo apt update && sudo apt install aicoder
Verify SHA-256
sha256sum aicoder_1.5.0_amd64.deb2d5fe933c48c440847dc29e34c3d4325a16bf7aded6aa972c2f1e58ae302a192
The calculated hash must match exactly.
Windows installer
50.6 MiB · Windows x64 · published September 29, 2026
Verify SHA-256
Get-FileHash .aicoder-1.5.0-setup.exe -Algorithm SHA2561fb53a0a1ee3256de860259d420c8059d1a6ece9dbaf17b83ac15b3b477f0437
The calculated hash must match exactly.
Source install for Android.
Android uses Bionic instead of glibc, so the normal Linux binary is not compatible with Termux. The Termux installer installs the required packages and Python dependencies, clones the current ai-coder source into ~/.local/lib/aicoder-src, and creates a local aicoder wrapper. No root access is required.
One-line install
curl -sL https://ailinux.me/downloads/AI-Coder/aicoder-1.5.0-termux-install.sh | bash
source ~/.bashrc
aicoderUpdate from source
cd ~/.local/lib/aicoder-src
git pullThe installer defaults to the current master source. Advanced users can set AICODER_REF before running it to select another branch or tag.
From login to autonomous agent.
aicoder login
aicoder # setup / interactive agent REPL
aicoder gui # PyQt6 desktop interface
aicoder models # available backend models
aicoder model <provider/model>
aicoder fallback <provider/model>
aicoder statusAn agent loop designed to finish the job.
Before you install
Does ai-coder run the model locally?
Not necessarily. TriForce can route inference to hosted providers while tool execution and system interaction remain on the AILinux machine.
Can I choose my own model?
Yes. ai-coder supports an explicit primary model and a separate fallback model from the TriForce catalog.
Does it only write code?
No. Depending on enabled tools and permissions, it can work with files, Git, logs, services, containers, diagnostics, search, and remote infrastructure.
Why 300 seconds by default?
Repository analysis, builds, debugging and long agent turns can legitimately take much longer than normal chat.
Give the agent a real task.
Install ai-coder 1.5.0, connect to TriForce, choose a model, and let the agent inspect, execute and verify work directly on AILinux.