bogware.com · Services · AI engineering assessment · Open source · Field notes · Start a project →
Bogware is a small engineering studio. We do cloud, platform and AI work for teams who could use a hand, and we publish the tools we make for ourselves as open source: Terraform providers, an LLM gateway, an agent harness, log search, and more. The READMEs are the ones we wished we'd had, and the issues get answered.
Like what you see in the code? We do the same kind of work for clients. Tell us what you need →
| Service | What it means for you |
|---|---|
| Custom AI chatbots | Assistants built on your own documents and data. They cite their sources, and when they don't know, they say so. |
| Coding-agent setup | Claude Code, Codex, Gemini CLI, Cursor, Copilot or bog-agents, set up for the way your team actually works. |
| AI engineering assessment | Two weeks: how your team builds today, and a written plan for working well with coding agents. Optional setup, or start with a one-week sprint. |
| Automation & RPA | The repetitive work goes to a machine that doesn't tire, with monitoring and a person to call. |
| Cloud, platform & Terraform | Infrastructure as code, custom Terraform providers, and platforms your developers won't route around. |
| AI governance & LLM gateways | One door for every model call: routing, cost limits, guardrails and an audit trail, in your own cloud. |
| Evaluation & LLM quality | Tests for prompts and agents, run in CI, so a model change is reviewed like a code change. |
| Log search & security analytics | Self-hosted search and detection over your own logs, without paying by the gigabyte. |
A written scope and a fixed price before work begins. What we build is yours: the code, the docs, the keys. → How an engagement goes
Made for ourselves first. Shared because it was useful. Every project has a page with install commands at bogware.com/work.
AI & agents
| Project | What it does | Stack |
|---|---|---|
| bog-agents | An agent framework: a Python SDK, a terminal coding agent, and a daemon for unattended runs. Hard cost caps, sandboxing, and a findings ledger you can gate CI on. | Python · MIT |
| gulley | One self-hosted gate for every model request: routing, budgets, guardrails, in-stream DLP and a tamper-evident audit trail. | TypeScript · Apache-2.0 |
pipx install bog-agents-cli # then run `bog-agents` in any repoData & security
| Project | What it does | Stack |
|---|---|---|
| cabinet | Self-hosted log search and security analytics. Plain-English questions, SPL when you want it. | C# · AGPL-3.0 |
| bql | A pipe query language and streaming engine for any row source. | C# · AGPL-3.0 |
| scriven | Run LangSmith evaluations on demand, over a WebSocket or as a gate in CI. | Python · MIT |
Infrastructure
| Project | What it does | Stack |
|---|---|---|
| bogform | Infrastructure as code without a hosted platform. A fork of OpenTofu: same language and workflow, no subscription. | Go · MPL-2.0 |
| terraform-provider-langsmith | LangSmith as code: projects, datasets, prompts, evaluators and access. | Go · MPL-2.0 |
| terraform-provider-unstructured | Unstructured document pipelines as code: sources, destinations and workflows. | Go · MPL-2.0 |
| terraform-provider-mergewf | Merge Workforce groups, tool access policies, skills and SCIM identities as code. | Go · MPL-2.0 |
| terraform-provider-mergeah | Merge Agent Handler tool packs, connectors, credentials and security rules as code. | Go · MPL-2.0 |
| terraform-provider-akamai | An independent Terraform provider for Akamai, separate from the official one. | Go · MPL-2.0 |
Audio & video
| Project | What it does | Stack |
|---|---|---|
| morass | A vocal channel strip plugin (AU, VST3, standalone): the whole chain on one screen. | C++ · GPL-3.0 |
| damp | A true-peak limiter and automated mastering engine, as a CLI, a GUI and a library. | C++ · AGPL-3.0 |
| moist | A video encoder and restoration suite that checks its own work. | C# · AGPL-3.0 |
In the workshop. We also run private tools for our own work, including an MCP gateway, an agent that fixes vulnerable dependencies and proves the fix, and retrieval over your own data. Clients can pilot them by arrangement. See the workshop →
- A person answers you, not a queue.
- If we're the wrong fit, we say so early and point you somewhere better.
- What we build for you is yours: the code, the docs, the keys.
- Calm software. No tricks, no traps, no lock-in.
- We finish what we start.
- Evolving Software: the tools have changed; the job of knowing what matters hasn't.
- Rethinking How We Build in the Generative Era: axioms, team shape and cadence for working with agents.
- All field notes → · RSS
| Start with | How |
|---|---|
| Request a quote | Four short steps. A written estimate within two working days. → bogware.com/contact |
| Book a call | Thirty minutes with an engineer. No salespeople on our side. → Pick a time |
| Send an RFP | RFPs, RFQs, tenders and security questionnaires. We confirm within a working day whether we'll respond. → bogware.com/contact/rfp |
| Just say hello | hello@bogware.com. Every message is read by a person. |
Found a bug or have an idea for one of the tools? Open an issue on its repository. They get answered.
A few engineers who like to help. No drama, no fuss.