Speech-to-text errors cluster in brittle terms: names, commands, flags, paths, identifiers, acronyms, versions, numbers, negation, and action words. Focus on a brittle term only when a wrong interpretation would change the action, target, scope, result, or safety. Accept ordinary wording as given.
Invocation
Read the argument text that follows the explicit skill invocation. Match the control words after trimming whitespace and without case sensitivity. Treat a control word as a control only when it is the entire argument.
on: Enable dictation mode for the current conversation. Confirm the new state. Apply the protocol to each later user message until the mode is disabled.off: Disable dictation mode. Confirm the new state. Treat later messages as normal text.- Any other non-empty argument: Apply the protocol to that request once. Preserve the current mode state. A later clarification is normal text when the mode is off.
- No argument: Report whether dictation mode is on or off, then state that this skill accepts
on,off, or a dictated request.
Initialize dictation mode as off in each new conversation. Store its state only in the
current conversation. Control invocations do not receive a Heard: readback.
Protocol
- Make a concise
Heard:readback the first text of the first assistant message for each dictated request, before any skill announcement, tool call, or task work. Restate the action, target, scope, and all consequential constraints. Use one line when it preserves them; otherwise use a short block.- Format each correction or assumption as code. Include the original transcript when the
change is not obvious:
ruff check(from "rough check"). - Mark an unresolved phrase and its credible candidates instead of choosing one:
[unclear: feature/cash-sync or feature/cache-sync?]. - Replace a secret or sensitive value with a role label such as
[API token].
- Format each correction or assumption as code. Include the original transcript when the
change is not obvious:
- Resolve every consequential brittle term before work that depends on it. Use the conversation, repository state, and cheap read-only checks to test candidates. Treat a term as resolved only when one candidate has direct support and no credible alternative remains.
- Proceed in the same turn when all dependent terms are resolved. The readback is an error check, not a confirmation gate.
- When two or more credible interpretations remain, ask one focused question before the dependent work. Complete every independent part of the request while that part is blocked. Mutate state only when the action and its target are both resolved.
The protocol is complete when every consequential ambiguity is resolved and stated, or is isolated behind a specific question, and all independent work is complete.