Repository navigation
feat(discord): respond in the user's message language #288
Copy link
Copy link
Closed
Labels
area:runtimeOpenClaw runtime and DockerOpenClaw runtime and Dockerarea:skillsGentle-AI/OpenClaw skillsGentle-AI/OpenClaw skillsenhancementNew feature or requestNew feature or requestpriority:p1Important but not immediately blockingImportant but not immediately blockingrisk:mediumMedium risk or dependency uncertaintyMedium risk or dependency uncertaintysize:SReview size <=300 changed linesReview size <=300 changed linesstatus:approvedApproved for implementationApproved for implementationtype:featureFeature workFeature work
Description
Pre-flight checks
- I searched existing issues and this is not a duplicate
- I understand this issue must be approved before implementation starts
- I will not include real Discord IDs, tokens, credentials, screenshots, raw logs, transcripts, or private payloads in public evidence
Problem
The OpenClaw Discord agent currently answers some Spanish user messages in English. This makes the tool feel inconsistent and harder to use for Spanish-speaking users, especially when the user writes in Spanish from Spain or another Spanish variant.
Commands and command names should remain in English for stability, but explanatory responses, guidance, clarification questions, and advisory text should adapt to the language used by the user in the current Discord message.
Proposed solution
Add a Discord response-language policy for OpenClaw-managed replies:
- Detect the language of the current user message at response time.
- Answer in the same language as the user's message by default.
- Keep commands, command names, code, config keys, channel identifiers, filenames, and copyable command snippets in English.
- If the message is mixed-language, prefer the user's dominant natural language while preserving technical terms.
- If confidence is low, ask a short clarification in the most likely language instead of silently falling back to English.
- Cover Spanish messages explicitly, including Spanish from Spain wording, without forcing regional slang.
Area
Runtime
Acceptance criteria
- A documented Discord response-language rule exists for OpenClaw assistant replies.
- Spanish user messages receive Spanish explanatory responses while command snippets remain in English.
- English user messages continue receiving English responses.
- Mixed Spanish/English messages preserve technical terms and command names without translating them incorrectly.
- The rule applies to general advisor/help responses and normal OpenClaw Discord responses.
- Fake-first fixtures cover Spanish, English, and mixed-language Discord prompts.
- No real Discord IDs, private transcripts, screenshots, or personal context are committed.
Validation plan
- Add sanitized fake Discord prompt/response fixtures.
- Validate that Spanish prompts produce Spanish prose and English command snippets.
- Validate that English prompts remain English.
- Review output for private-data leakage and over-translation of commands/config.
Reactions are currently unavailable
Metadata
Metadata
Assignees
Labels
area:runtimeOpenClaw runtime and DockerOpenClaw runtime and Dockerarea:skillsGentle-AI/OpenClaw skillsGentle-AI/OpenClaw skillsenhancementNew feature or requestNew feature or requestpriority:p1Important but not immediately blockingImportant but not immediately blockingrisk:mediumMedium risk or dependency uncertaintyMedium risk or dependency uncertaintysize:SReview size <=300 changed linesReview size <=300 changed linesstatus:approvedApproved for implementationApproved for implementationtype:featureFeature workFeature work
Activity
egdev6 commentedon Jul 22, 2026
Completed by PR #290: Discord/OpenClaw response-language policy merged to develop. Closing manually because the PR targeted develop.
egdev6 commentedon Jul 24, 2026
Post-release hygiene: v0.5.0 has been promoted to
mainvia PR #296 and GitHub Releasev0.5.0.Project status updated to
Merged to main.No production/public Discord/live publishing/live analytics readiness is claimed; the release remains internal fake-first/local plus private Discord no-op verification where applicable.