fix(doctor): MiniMax /v1/models returns 404 by design — check key presence only#818
Closed
ygd58 wants to merge 1 commit into
Closed
fix(doctor): MiniMax /v1/models returns 404 by design — check key presence only#818ygd58 wants to merge 1 commit into
ygd58 wants to merge 1 commit into
Conversation
MiniMax global API does not support the /v1/models endpoint, causing hermes doctor to always report HTTP 404 even when the key is valid and the provider works correctly in chat sessions. Fix: remove MiniMax from the endpoint-check list and show it as configured (key-only check) to avoid the misleading warning. Fixes NousResearch#811
1 task
Contributor
Author
|
Closing in favor of #833 which takes a more general approach — adds a check_url flag to the provider tuple so any future provider with the same limitation only needs a single field change. Also fixes MiniMax (China) which had the same unreported issue. |
lmsanch
added a commit
to lmsanch/hermes-agent
that referenced
this pull request
Apr 22, 2026
…esearch#818) Follow-up to d4fc986 — the telegram.py edit didn't make it into that commit due to a linter race. This adds the missing import + send() integration. Closes lmsanch/toryx-private#818 (replaces the partial close from d4fc986)
lmsanch
added a commit
to lmsanch/hermes-agent
that referenced
this pull request
Apr 22, 2026
…sResearch#818) Follow-up to f0e63ec extending the redaction chokepoint beyond telegram and email: - tools/tts_tool.py::text_to_speech_tool: redact `text` right after the empty-string guard. Every TTS provider (edge/elevenlabs/openai/fish/ minimax/mistral/gemini/neutts) receives already-redacted input, so voice output can't leak secrets even though text output is fine. - gateway/platforms/discord.py::send: same pattern as telegram/email — redact `content` at the top of the method. Remaining platform adapters (slack, feishu, wecom, dingtalk, mattermost, signal, bluebubbles, homeassistant, matrix, qqbot, sms) follow the same one-line pattern and can land in a follow-up PR. Still refs lmsanch/toryx-private#818
lmsanch
added a commit
to lmsanch/hermes-agent
that referenced
this pull request
Apr 22, 2026
…ation (#4) * feat(security): secret-redaction preprocessor + telegram/email integration Motivating incident: toryx-private#807 Issue 5 — Christopher leaked raw Alpaca API keys into a Telegram approval prompt. This module closes that class at the send-path chokepoint. New `tools/secret_redactor.py`: - Regex-based redaction for common provider keys (Anthropic, OpenAI, Groq, Fireworks, Toryx, GitHub PAT, Slack, AWS) + Alpaca labeled headers + PEM private keys + generic Bearer tokens - Replacement marker: [REDACTED:<kind>] so readers can see what was stripped - Idempotent, fast (precompiled patterns), safe on None/empty input Integration: - gateway/platforms/telegram.py send(): redact content at the top of the method, before format_message() and chunking - gateway/platforms/email.py send(): redact content before loop.run_in_executor passes it to _send_email All test vectors use synthetic A/B/1 fillers to avoid tripping push-protection secret-scanning. No real keys in tests. Not yet wired (follow-up PR): TTS text at tts_tool.py::_generate_fish_audio, plus the other platform adapters (discord, slack, feishu, wecom, etc.). Closes lmsanch/toryx-private#818 * feat(security): wire secret redactor into telegram adapter too (NousResearch#818) Follow-up to d4fc986 — the telegram.py edit didn't make it into that commit due to a linter race. This adds the missing import + send() integration. Closes lmsanch/toryx-private#818 (replaces the partial close from d4fc986) * feat(security): wire secret redactor into TTS + discord adapters (NousResearch#818) Follow-up to f0e63ec extending the redaction chokepoint beyond telegram and email: - tools/tts_tool.py::text_to_speech_tool: redact `text` right after the empty-string guard. Every TTS provider (edge/elevenlabs/openai/fish/ minimax/mistral/gemini/neutts) receives already-redacted input, so voice output can't leak secrets even though text output is fine. - gateway/platforms/discord.py::send: same pattern as telegram/email — redact `content` at the top of the method. Remaining platform adapters (slack, feishu, wecom, dingtalk, mattermost, signal, bluebubbles, homeassistant, matrix, qqbot, sms) follow the same one-line pattern and can land in a follow-up PR. Still refs lmsanch/toryx-private#818
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
MiniMax global API does not support the
/v1/modelsendpoint, causinghermes doctorto always report⚠ (HTTP 404)even when the API keyis valid and MiniMax works correctly in chat sessions.
Fix: remove MiniMax from the endpoint-check list and show
✓ (API key configured)when
MINIMAX_API_KEYis present — consistent with how other key-onlyproviders are handled.
MiniMax (China) is unaffected — it keeps its existing endpoint check.
Fixes #811