Ship faster with context-aware AI workflows.

BitsentryAI gives your OpenCode agent structured SDD, SDR, and Security profiles, so it plans, researches, and executes with the right context and guardrails.

Download / Install How to use 
# Ask naturally
$ opencode "let's improve onboarding"

# Bitsentry routes by intent
# SDD → spec/design/apply discipline
# SDR → research/evidence first

# Result: better decisions, safer execution

Without context, AI agents make expensive mistakes.

  • 01 Teams mix coding, research, and security tasks in one generic prompt flow.
  • 02 Important decisions and constraints get lost between sessions.
  • 03 Lack of explicit guardrails increases risk and rework. 
# Same tool, different intents, zero structure

# Build feature + security check + research note
# all requested in one unbounded execution

# No workflow discipline
# No persistent context memory

# Result: inconsistent output and avoidable risk

One product. Clear operating modes.

What it is: A context-aware orchestration layer for OpenCode. BitsentryAI matches intent to the right workflow profile, applies guardrails, and keeps context persistent.

⚡

SDD Profile

For product changes. Spec, design, acceptance criteria, and implementation discipline before edits.

🔍

SDR Profile

For discovery and diagnosis. Hypotheses, evidence, and non-mutating investigation first.

🛟

Security Profile

For controlled security context. Scope, authorization, and explicit approvals before sensitive actions.

Build spec-driven product delivery
Investigate research, diagnosis, discovery
Protect scope-gated security reviews

Backed by Engram (persistent memory), Context7 (fresh docs), and native MCP integration.

Why teams use it: less context switching, fewer wrong actions, safer execution, and better cross-team traceability.

Current compatibility

○

OpenCode

Primary and validated in this MVP

Primary
○

Claude Code

Anthropic's CLI agent

Planned
○

Cursor

AI-first code editor

Planned
○

Codex

OpenAI's CLI agent

Planned

Context that makes decisions better

Three context layers reduce mistakes and improve consistency across sessions.

Engram

Remembers decisions, constraints, and handoffs across sessions.

Active

Context7

Keeps framework and library docs fresh while your agent works.

Active

Readiness checks

Shows setup status before your team depends on external context tools.

Built-in

How it works in practice

1

Install and bootstrap

Run the installer once and configure BitsentryAI for OpenCode with guided defaults.

$ ./install.sh
$ bitsentry-ai

# Open Install/Setup in TUI
2

Choose the right profile by intent

BitsentryAI routes work into SDD, SDR, or Security profile based on your goal.

# SDD: build/change product
# SDR: investigate/research
# Security: scoped, controlled review
3

Execute with guardrails

Each profile enforces different outputs and risk gates so actions stay consistent and auditable.

$ opencode "build a feature"

# Route preview → plan/spec/design → apply
4

Keep context between sessions

Decisions, discoveries, and constraints persist, so the next session starts informed instead of from zero.

# Engram stores what was decided
# Context7 keeps documentation fresh

Roadmap and transparency

OpenCode is the current focus. Additional agent support is planned and tracked publicly. We build in the open with explicit scope boundaries.

Follow the progress → GitHub
Now

OpenCode Support

Full SDD, SDR, Support flow integration

Q2 2026

Claude Code Integration

MCP protocol support

Q3 2026

Cursor + Codex

IDE agent capabilities

Install BitsentryAI and give your agent the right context.

Download from GitHub → Installation guide

Public MVP: controlled manual workflows, explicit readiness, no hidden autonomous runtime.