Develop - #434
Merged
Merged
Conversation
* Require storage in stack resource validation * Reject invalid stack storage values * chore: inc ver
* feat: add CyberSecQwen serving engine What changed: - added a dedicated cybersec_qwen_4b AI engine mapping - added a CPU-bounded CyberSecQwen Q4_K_M llama.cpp serving profile - added a lightweight registry/config smoke test that does not require llama_cpp Why: - give RedMesh a stable local GGUF engine target without relying on a generic llama_cpp alias * feat: route RedMesh LLM through local provider What changed: - made REDMESH_LLM_AGENT_API provider-aware with local LLM_INFERENCE_API as the default - kept DeepSeek as an explicit opt-in provider - normalized local LLM responses to the existing /chat and /analyze_scan contracts - added redacted local health/status handling and provider-selection tests Why: - replace hidden remote LLM dependency with local CyberSecQwen-ready serving while preserving RedMesh operator workflows * test: add local RedMesh LLM provider e2e What changed: - added a loopback e2e-style smoke for REDMESH_LLM_AGENT_API local provider routing - validates health redaction, chat, analyze_scan, max-token clamping, and no Authorization header when no local token is configured Why: - prove the local LLM provider path without requiring GGUF weights or external network access * docs: fix local inference engine examples What changed: - replaced invalid generic llama_cpp example engine names - pointed the LLM_INFERENCE_API example at the dedicated cybersec_qwen_4b engine Why: - keep operator examples aligned with the actual AI engine registry * style: trim CyberSecQwen serving file What changed: - removed the extra blank line at EOF from the CyberSecQwen serving process Why: - keep the edge-node branch passing whitespace validation * fix: harden local LLM status metadata What changed: - dropped query and fragment data from LLM health URLs - counted local and remote config errors consistently in status metrics - defaulted structured-report model provenance to CyberSecQwen while preserving overrides - added focused regression tests for redaction, metrics, and model provenance Why: - prevent health/status leaks and stale DeepSeek provenance in local RedMesh LLM mode * fix: support mounted llama.cpp model files * fix: bound RedMesh local LLM report prompts * chore: include sqlfluff in devcontainer requirements * feat: add provider-specific RedMesh LLM prompt profiles * refactor: use generic remote LLM model config * refactor: use generic remote LLM provider config What changed: - replace public REDMESH_LLM_AGENT_API DeepSeek-specific config keys with REMOTE_LLM_* keys - make LLM_PROVIDER=remote the explicit remote mode and keep deepseek as the adapter value - update provider and prompt-profile tests for the generic remote path Why: - remove provider-specific public config names while preserving local CyberSecQwen behavior * fix(red_mesh): route auto provider through model-based inference for prompt profile The structured-report adapter pre-resolved an auto PROMPT_PROFILE against a literal remote whitelist (remote/openai/anthropic). With PROVIDER=auto that fell to the local branch and handed resolve_prompt_profile a concrete profile name, short-circuiting the model-based inference. A remote model (e.g. deepseek-chat) then received the local json_schema response_format instead of json_object. Resolve the auto profile through infer_provider_path so an auto provider picks the remote profile when the model itself is remote. Add a regression test covering PROVIDER=auto + deepseek-chat. * chore: inc ver
* Migrate TCP tunnels to port registry * Use plugin RNG for TCP port allocation * Remove tunnel manager safe log wrapper * Sync TCP routes from tunnel manager process * Drop TCP route created timestamp * Attach TCP routes by metadata port hint * Return TCP display fields outside metadata * Omit empty TCP alias public id * Sample TCP public ports without shuffling * Address TCP tunnel PR review comments * chore: inc ver * Reduce default TCP public port range * Harden TCP route claim cleanup
aledefra
approved these changes
Jun 5, 2026
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
3 participants
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.
No description provided.