Skip to content

Decide before Speak.

decidespeak is a harness that reads, edits and runs commands in your project, and calls MCP tools. Every action clears a deterministic gate first, and nothing counts as done without evidence.

It decides before it acts.

Every action meets the same consequence check. It is deterministic code, not a model, and it answers allow, confirm or deny before anything runs. Approve mode clears low-risk actions on its own; Strict asks for every one.

allow

["ls", "-la", "src/"]

Read-only. It goes straight through.

confirm

["rm", "-rf", "build/"]

Destructive. It waits: allow once, allow from now on, or refuse.

deny

["curl", "https://api.example.com"]

Your deny rule matches. Nothing can loosen it.

Deny wins. Allow, ask and deny rules are created only by you, and no setting, mode or model output can loosen a deny. Commands are argv arrays, never shell strings.

The mark

Six frames of one sentence.

The logo is the product, drawn. The D is the gate and it does not move. The S is the action, and it only gets anywhere by passing through.

Before Nothing has been decided. Nothing runs.
Decide The gate settles first. It is code, not a model.
Propose An action sets off along the S, toward the D.
Allow Cleared. It passes through and carries on.
Confirm It holds at the gate until you answer.
Deny It stops at the D. Nothing loosens that.

How it works

It starts from no, and has to be let through.

confirm

Cards show the whole action

A confirm card shows the full content. Allow once, allow from now on, or refuse. Change the action and the old confirmation is void.

undo

Snapshot before overwrite

A file is snapshotted before it is overwritten, and the write clears on its own only if the snapshot succeeded. Edits can be undone.

sandbox

Commands run confined

On macOS, and on Linux with Landlock, writes stay inside the project and the network is denied. Reads are not confined. Where no sandbox exists, every command asks, and an unavailable sandbox is an error, never a silent downgrade.

verify

Done means evidenced

Give it acceptance criteria and a turn needs deterministic checks plus an independent review bound to the artifact revision. Without criteria, the reply is still fact-checked against the record.

scope

Subagents on a leash

Task subagents get a required file scope, and the gate refuses edits outside it. Read-only ones run up to four at once. No subagent verifies its own work.

mcp

Tools behind the same door

MCP servers are managed in the app. Every call runs behind schema validation, the gate, approval, a timeout and output limits.

Modes

One loop, three ways to run it.

Mode is separate from the permission mode. There is no chat-only mode: small talk is just the loop's first reply.

auto

Default

The main loop. Each round it decides whether the situation calls for the decision layer on top.

plan

Plan only

The same loop, read-only: the gate refuses writes and MCP calls. It ends on a plan card. Run it, keep planning, or throw it away.

decide

Decision loop

The decision layer on every round: candidate directions, a routed choice, a coder subagent and a fresh-context reviewer.

A turn

A turn, end to end.

Rounds, calls, time, failures and lack of progress all stop a task as paused or failed. Never as completed.

rounds
4
calls
12
seconds
600
stalls
2

Defaults. Each is a setting.

It proposes

Each step the model returns a batch of actions, a reply, or a question. A message you send mid-turn is picked up at the next step. Long conversations are compacted deterministically, and large results go to disk.

The gate clears it

Read-only actions run in parallel; everything else runs one at a time, each through the gate. A command that runs past 30 seconds moves to the background, with live output and a cancel button.

It gets checked

Completion goes through deterministic checks, an independent review and an omitted-evidence check. A reply that claims a file changed when it did not is sent back, then replaced by an answer written from the record.

Or it doesn't finish

Only passing counts as done. Anything else goes back into the loop with the reasons attached. The event log is the only record, so a restart or a refresh restores every task.

Decision layer

Underneath, a model that only decides.

Laya is an open-source, non-generative model: a state and typed questions in, every answer with a probability out, in one forward pass. Jev, hosted by TypeSafe, answers the same questions over the network.

decide() — one call, every answer
{
  "type": "choice",
  "choice": "refactor first",
  "probabilities": {
    "refactor first": 0.71,
    "patch in place": 0.22,
    "ask the user":  0.07
  },
  "confidence": 0.71,
  "action": { "act_probability": 0.68 }
}

Low confidence means it asks

Below the calibrated threshold (0.55 by default) you get a question, not a guess.

Two backends, one contract

Registering another backend is a line in a table. The chat model is pluggable too: local Ollama, or any OpenAI-compatible server.

Honest about its edges

Laya reads 512 tokens (1024 multilingual) and truncates from the end. Choice questions degrade past roughly twenty options.

Manifesto

Most agents speak first and explain later. This one decides first, and speaks only once it is through.

  • i. No is the default.
  • ii. Deny always wins.
  • iii. Done needs evidence.

Bring an agent that knows when to stop.

Access is limited. Sign in to get the build and the documentation.

  • Runs on your machine
  • Desktop app · macOS, Apple silicon
  • Linux and Windows · from source