Spec-driven
Numbered Markdown specs are the source of truth. Schema-validated, git-diffable, and the only input the build consumes.
Evolve Verified Software
Overplane combines popular AI coding agents with containers, spec-driven development, and formal verification to create a safer and more reliable AI coding workflow.
Free & open source · Apache-2.0 licensed
Safer code Auditable codegen Model agnostic
The prop
Coding agents earned their reputation for speed, and for the occasional catastrophe. Overplane keeps the speed and fixes the rest in three moves: contain the agent, wrap it in a real build system, and verify the plan before the code exists.
One agent deleted a production database during a code freeze. Another destroyed a user’s files while “reorganizing” them. Even the agent vendors now say unattended runs belong in a container behind an egress firewall.
The fix
In Overplane the container is not a best practice you set up later. It is the only way agents run: every agent, every build, the same locked-down container with a restricted view of your system. A bad day means deleting a sandbox, not restoring a backup.
Most spec-driven tools are Markdown prompts riding inside your coding agent. Run one twice and you get two different codebases, with no record of which run produced what. There is a growing case that code generation now sits in the trusted computing base , where determinism and auditability stop being optional.
The fix
Overplane treats codegen like compilation: content-hashed sandbox images, content-addressable outputs, cached incremental builds you can replay down to a single spec, and a normalized dollar cost for every agent run. A real build system, where the compiler happens to be an agent.
96% of developers don’t fully trust AI-generated code, and over a third say reviewing it takes more effort than reviewing a colleague’s. Generated diffs are the new toil.
The fix
Review moves upstream, where it is cheap. You sign off on short numbered specs, Z3 checks them for contradictions before any code is written, and the checked IR can keep going: in the rustdis example it becomes 48 passing property tests and Kani proof harnesses guarding a wire-compatible Redis clone.
What is it?
Overplane is a small, single-file, open-source binary that orchestrates
recipes, prompts, containers, and AI coding agents into an opinionated
workflow:
plain-language specs in, working and verified software out.
Numbered Markdown specs are the source of truth. Schema-validated, git-diffable, and the only input the build consumes.
Every coding agent runs in an identical local container with a restricted view of your system. OS-level isolation ensures reproducibility and security.
Content-hashed images and content-addressable outputs make builds cacheable, auditable, and replayable down to an individual spec.
Raise lifts each spec into an Intermediate Representation; verify runs Z3 on per-spec and merged models before any code is written.
Codegen gives the agent both the checked IR and the original spec, then reconciles the generated output only after a successful run.
Use your own API keys with Claude Code, Codex, or OpenCode, and choose the agent per project or per run.
Multi-agent support
Overplane is not another agent, and it is not another subscription. It drives the headless agents you already pay for, on your own API keys, and lets you swap them per project or per run.
Every agent below installs from official channels into a container image you build, audit, and control.
Main features
The closest cousin is Spec Kit , and it is genuinely good: prompts that bring spec discipline to the agent you already run. The difference is what happens after the spec. Spec Kit hands your agent a better prompt; Overplane runs the whole build, and owns the sandbox, the cache, the verification step, and the bill.
| Capability | Overplane | Spec Kit 0.12 | Kiro IDE 1.0 / CLI 2.12 |
|---|---|---|---|
Spec-driven authoringNumbered, schema-validated specs are the durable source of truth for what you are building. | Yes | Yes | Yes |
Container-isolated agentsEvery AI agent runs sandboxed in its own container, never touching your host. | Yes | No | No |
Vendor-neutral multi-agentOne abstraction over Claude Code, Codex, and OpenCode. Pick your agent per project or per run. | Yes | Yes | No |
Unified cost reportingNormalized token usage and dollar cost for every agent run, aggregated automatically per build. | Yes | No | No |
Cached, hashed buildsDeterministic project images and content-addressable file sets make runs auditable and repeatable. | Yes | No | No |
Spec verificationAutomated consistency and validity checks on specs before code is generated. | Yes | Yes | Yes |
Form factorHow the tool ships, and where the workflow actually runs. | One CLI binary | Prompts in your agent | IDE + CLI + web |
Open source licenseThe license under which the tool itself is distributed. | Apache-2.0 | MIT | Proprietary |
Peer snapshot (July 2026, Spec Kit 0.12.x, Kiro IDE 1.0.x / CLI 2.12.x):
Overplane validates specs against a schema, then runs
raise and verify with Z3 consistency and scenario
checks in overplane build. Spec Kit ships
/speckit.clarify, /speckit.analyze, and
/speckit.checklist quality gates plus
/speckit.converge for drift; Kiro generates property-based and
unit tests from its specs. Approaches differ; all three verify specs before
implementation. On multi-agent: Spec Kit targets 30+ agents; Kiro offers many
models (Claude and open-weight) inside its own agent, which is not the same
thing. On isolation: Kiro's hosted web sessions run in cloud sandboxes, but
the IDE and CLI run agents directly on your machine. On cost: Overplane normalizes
token usage and USD cost across agent backends per build; Spec Kit has no built-in
cost tracking; Kiro bills in plan credits with account-level dashboards, not
per-run dollar cost.
Install the CLI and run your first spec-driven build in minutes.