===== PHASE 02 CAS QWEN LOCAL CAPACITY AUDIT ===== PHASE_ID: phase-02 CATEGORY_ID: architecture_engineering_core DOC_ID: cas DOCS: /home/yeff/public_html/devon/docs/index.php DATA: /home/yeff/public_html/devon/panel/data CANON: /home/yeff/public_html/devon/canon OUT: /home/yeff/public_html/devon/_audit/phase02_cas_qwen_local_capacity_audit_20260429_092427.txt ===== SOURCE OF TRUTH CHECK ===== PASS: docs/index.php present PASS: panel/data present PASS: canon present PASS: hub_index.json present ===== PHP SYNTAX CHECK ===== No syntax errors detected in /home/yeff/public_html/devon/docs/index.php ===== CONTEXT DUMP SUMMARY ===== CATEGORY: architecture_engineering_core | LABEL: Architecture & Engineering Canon | PHASE: phase-02 | DOCS: 18 ===== DOCS INDEX forceDirectArchitectureRender ===== const forceDirectArchitectureRender = - master_architecture_index: CUSTOM_BRANCH - panel_manifest: CUSTOM_BRANCH - project_scope: CUSTOM_BRANCH - deployment_order: CUSTOM_BRANCH - sandbox_environment: CUSTOM_BRANCH - server_registry: CUSTOM_BRANCH - project_progress_canonical: CUSTOM_BRANCH - project_progress_model: CUSTOM_BRANCH PHASE-02: CATEGORY: architecture_engineering_core | Architecture & Engineering Canon - cas: CUSTOM_BRANCH - cgs: CUSTOM_BRANCH - acs: CUSTOM_BRANCH - css: CUSTOM_BRANCH - ncs: CUSTOM_BRANCH - sts: CUSTOM_BRANCH - brps: CUSTOM_BRANCH - cdms: CUSTOM_BRANCH - ris: CUSTOM_BRANCH - ofms: CUSTOM_BRANCH - card_contract_minimums: CUSTOM_BRANCH - panel_data_contract: CUSTOM_BRANCH - panel_component_contract: CUSTOM_BRANCH - panel_sync_contract: CUSTOM_BRANCH - panel_canonical_tree: CUSTOM_BRANCH - panel_navigation_spec: CUSTOM_BRANCH - panel_ui_blueprint: CUSTOM_BRANCH - canonical_matrix_v1: CUSTOM_BRANCH No syntax errors detected in /home/yeff/public_html/devon/docs/index.php ===== DUMP STATUS ===== PASS: canonical + docs + hub context reconstructed from server evidence ===== HOST IDENTITY ===== Waresites Linux Waresites 5.15.0-176-generic #186-Ubuntu SMP Fri Mar 13 11:01:42 UTC 2026 x86_64 x86_64 x86_64 GNU/Linux Distributor ID: Ubuntu Description: Ubuntu 22.04.5 LTS Release: 22.04 Codename: jammy ===== CPU ===== Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Address sizes: 48 bits physical, 48 bits virtual Byte Order: Little Endian CPU(s): 6 On-line CPU(s) list: 0-5 Vendor ID: AuthenticAMD Model name: AMD EPYC Processor (with IBPB) CPU family: 23 Model: 1 Thread(s) per core: 1 Core(s) per socket: 6 Socket(s): 1 Stepping: 2 BogoMIPS: 5589.49 Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm rep_good nopl cpuid extd_apicid tsc_known_freq pni pclmulqdq ssse3 fma cx16 sse4_1 sse4_2 x2apic movbe popcnt aes xsave avx f16c rdrand hypervisor lahf_lm cmp_legacy cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw topoext ssbd ibpb stibp vmmcall fsgsbase bmi1 avx2 smep bmi2 rdseed adx smap clflushopt sha_ni xsaveopt xsavec xgetbv1 arat Hypervisor vendor: KVM Virtualization type: full L1d cache: 192 KiB (6 instances) L1i cache: 384 KiB (6 instances) L2 cache: 3 MiB (6 instances) L3 cache: 8 MiB (1 instance) NUMA node(s): 1 NUMA node0 CPU(s): 0-5 Vulnerability Gather data sampling: Not affected Vulnerability Indirect target selection: Not affected Vulnerability Itlb multihit: Not affected Vulnerability L1tf: Not affected Vulnerability Mds: Not affected Vulnerability Meltdown: Not affected Vulnerability Mmio stale data: Not affected Vulnerability Reg file data sampling: Not affected Vulnerability Retbleed: Mitigation; untrained return thunk; SMT disabled Vulnerability Spec rstack overflow: Mitigation; SMT disabled Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl and seccomp Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization Vulnerability Spectre v2: Mitigation; Retpolines; IBPB conditional; STIBP disabled; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected Vulnerability Srbds: Not affected Vulnerability Tsa: Not affected Vulnerability Tsx async abort: Not affected Vulnerability Vmscape: Not affected CPU_CORES_LOGICAL: 6 ===== MEMORY ===== total used free shared buff/cache available Mem: 11Gi 964Mi 9.8Gi 28Mi 990Mi 10Gi Swap: 0B 0B 0B MemTotal: 12243772 kB MemAvailable: 10896372 kB SwapTotal: 0 kB SwapFree: 0 kB ===== DISK ===== Filesystem Size Used Avail Use% Mounted on /dev/sda1 97G 25G 73G 26% / /dev/sda1 97G 25G 73G 26% / /dev/sda1 97G 25G 73G 26% / /dev/sda1 97G 25G 73G 26% / ===== GPU / NVIDIA / ACCELERATION ===== MISSING: nvidia-smi absent ----- lspci GPU scan ----- 00:00.0 Host bridge: Intel Corporation 440FX - 82441FX PMC [Natoma] (rev 02) 00:01.0 ISA bridge: Intel Corporation 82371SB PIIX3 ISA [Natoma/Triton II] 00:01.1 IDE interface: Intel Corporation 82371SB PIIX3 IDE [Natoma/Triton II] 00:01.2 USB controller: Intel Corporation 82371SB PIIX3 USB [Natoma/Triton II] (rev 01) 00:01.3 Bridge: Intel Corporation 82371AB/EB/MB PIIX4 ACPI (rev 03) 00:02.0 VGA compatible controller: Device 1234:1111 (rev 02) ===== LOCAL AI RUNTIME TOOLING ===== PASS: python3 => /usr/bin/python3 Python 3.10.12 MISSING: python PASS: pip3 => /usr/bin/pip3 pip 22.0.2 from /usr/lib/python3/dist-packages/pip (python 3.10) PASS: pip => /usr/bin/pip pip 22.0.2 from /usr/lib/python3/dist-packages/pip (python 3.10) MISSING: docker MISSING: docker-compose MISSING: ollama MISSING: llama-server MISSING: llama.cpp MISSING: llama-cli MISSING: llama-cpp-python MISSING: node MISSING: npm PASS: git => /usr/local/cpanel/3rdparty/lib/path-bin/git git version 2.48.2 PASS: curl => /usr/bin/curl curl 7.81.0 (x86_64-pc-linux-gnu) libcurl/7.81.0 OpenSSL/3.0.2 zlib/1.2.11 brotli/1.0.9 zstd/1.4.8 libidn2/2.3.2 libpsl/0.21.0 (+libidn2/2.3.2) libssh/0.9.6/openssl/zlib nghttp2/1.43.0 librtmp/2.3 OpenLDAP/2.5.19 Release-Date: 2022-01-05 Protocols: dict file ftp ftps gopher gophers http https imap imaps ldap ldaps mqtt pop3 pop3s rtmp rtsp scp sftp smb smbs smtp smtps telnet tftp PASS: wget => /usr/bin/wget GNU Wget 1.21.2 built on linux-gnu. -cares +digest -gpgme +https +ipv6 +iri +large-file -metalink +nls ===== PYTHON PACKAGES RELATED TO LOCAL LLM ===== MISSING: python module torch MISSING: python module transformers MISSING: python module accelerate MISSING: python module sentencepiece MISSING: python module tokenizers MISSING: python module llama_cpp MISSING: python module ctransformers MISSING: python module vllm MISSING: python module bitsandbytes MISSING: python module numpy ===== QWEN LOCAL MODEL SCAN ===== Searching common model/runtime paths. This may take a few seconds. ----- SCAN: /home/yeff/public_html/devon ----- /home/yeff/public_html/devon/_audit/phase02_cas_qwen_local_capacity_audit_20260429_092427.txt ----- SCAN: /home/yeff/public_html/devon/panel/data ----- ----- SCAN: /home/yeff/public_html/devon/canon ----- MISSING_DIR: /opt/devon ----- SCAN: /opt ----- ----- SCAN: /home/yeff ----- /home/yeff/public_html/devon/_audit/phase02_cas_qwen_local_capacity_audit_20260429_092427.txt MISSING_DIR: /home/yeff/.cache ----- SCAN: /root/.cache ----- MISSING_DIR: /var/lib/ollama ----- SCAN: /usr/local/share ----- MISSING_DIR: /models ===== OLLAMA MODEL LIST ===== MISSING: ollama absent ===== CAS BRANCH CURRENT API / QWEN / AI CLASS SCAN ===== CAS_BRANCH_SIZE_CHARS: 86657 TERM_SCAN: MISSING: qwen PASS: api => 38 PASS: external api => 10 PASS: openai => 1 MISSING: gpt PASS: claude => 1 PASS: gemini => 1 PASS: ollama => 1 MISSING: llama.cpp MISSING: gguf MISSING: transformers MISSING: torch PASS: local => 5 MISSING: self-controlled MISSING: model runtime PASS: reasoning => 13 PASS: planning => 8 PASS: memory => 18 PASS: retrieval => 16 PASS: embedding => 12 PASS: embeddings => 12 PASS: vector => 15 PASS: tool execution => 13 PASS: orchestration => 12 PASS: context => 21 API_VENDOR_RISK_LINES: 284: CAS must declare the AI architecture classes Devon needs before any cognitive boundary is trusted. The required layer is not an external API and not a vendor stack. CAS requires a future local/model execution layer, reasoning and planning, prompt/context manag 290: - local_or_self_controlled_model_runtime: required class; no external API dependency declared 302: - PASS: DH declares AI capability classes without naming vendor/API as decided 304: - FAIL: CAS names external API, vendor or library as required without server evidence or canonical decision 308: - external API is introduced as default architecture 309: - vendor/library is named without evidence 481: CAS configuration must separate AI capability classes from implementation products. It governs the need for reasoning, planning, memory, retrieval, embeddings, vector state, tool execution, orchestration and context management. It does not select OpenAI, Claud 488: - not_governed_by_CAS_yet: vendor, hosted API, external model provider, concrete vector database, framework library and runtime product 494: - PASS: CAS declares capability classes and explicitly blocks vendor/API inference 499: - external API becomes implied dependency 502: - JSON mirror later records vendor choice not present in DH 672: CAS evidence must distinguish what is already visible from what is not decided. The server evidence shows AI capability classes such as memory, embeddings, retrieval, vector state, reasoning, planner, prompt interpretation, tool execution, orchestration and co 675: This distinction is critical. CAS can declare architectural classes because the project already references cognition, memory, retrieval, embeddings, planner and tool execution domains. CAS cannot declare an API or product stack because the evidence does not su 687: - evidence_missing_product: external API provider, model vendor, AI framework, vector database implementation 692: - FAIL: evidence claim promotes vendor/API/library without canonical server proof 697: - API dependency is introduced because no local implementation is decided yet 865: title: "CAS Promotion Must Block Mirror Until AI Capability Classes Are Declared Without API Assumption", 868: Completion & Promotion must block CAS mirror generation until the DH declares AI capability classes clearly and without external API assumption. CAS may promote the requirement for local/model execution class, reasoning, planning, memory, retrieval, embeddings 876: - external_API_dependency: not declared 877: - vendor_or_library_selection: MISSING until future canonical decision 883: - PASS: CAS DH declares AI classes and blocks vendor/API inference 884: - FAIL: JSON mirror records external API, vendor or library before DH/canon decision 889: - CAS promotes API usage despite project rule 1065: title: "CAS Validation Must Check AI Class Declaration And Block API/Vendor Inference", 1068: Validation must check that CAS declares IA capability classes and does not smuggle in API/vendor assumptions. PHP syntax and hub binding still matter as DH substrate, but the CAS-specific validation target is whether the cognitive architecture names the requir 1077: - validation_command: grep CAS DH branch to ensure no API/vendor is treated as selected 1085: - FAIL: CAS declares vendor/API/library as chosen without evidence 1090: - API dependency introduced by wording 1257: title: "CAS Installation Must Install IA Capability Classes, Not API Stack, Inside Existing Buckets", 1263: The project is building IA Devon, not just a documentation hub. CAS installation must therefore expose the cognitive technologies the future Panel must design around. At the same time, it must avoid pretending that an API, vendor, framework or database product 1274: - forbidden_installation: external API as default, vendor stack, new Technology bucket 1279: - FAIL: requirements appear as vendor/API stack or separate Technology bucket 1284: - API stack appears as default 1475: title: "CAS Recovery Must Remove API Assumptions And Restore Capability-Class Authority", 1478: Recovery must remove any API, vendor, library or product assumption that enters CAS without evidence. The recovery path is to restore CAS as the DH authority for IA capability classes, then regenerate later JSON mirrors only from the corrected DH content. 1481: A false AI stack decision upstream is expensive. It can distort CFC, OFMS, memory, retrieval, tool execution and future Panel design. Recovery must therefore protect the category from two opposite failures: reducing CAS to PHP/JS/JSON substrate, or overcorrect 1487: - required_block: external API/vendor/library inference 1493: - FAIL: recovery edits JSON or downstream consumers while CAS still contains API/vendor assumptions 1497: - API dependency survives as implied requirement 1505: - remove or block API/vendor/library requirements not declared by evidence QWEN_LINES: CURRENT TECHNOLOGY CARD TITLES: - CAS Requires Local AI Architecture Classes Before Cognitive Boundaries Are Trusted - CAS Configuration Must Separate AI Capability Classes From Product Choices - CAS Evidence Must Distinguish Existing AI Class Evidence From Undecided Stack - CAS Promotion Must Block Mirror Until AI Capability Classes Are Declared Without API Assumption - CAS Validation Must Check AI Class Declaration And Block API/Vendor Inference - CAS Installation Must Install IA Capability Classes, Not API Stack, Inside Existing Buckets - CAS Recovery Must Remove API Assumptions And Restore Capability-Class Authority ===== PANEL DATA / CANON QWEN AND API SCAN ===== ----- Qwen references ----- ----- API/vendor references that may conflict with no-API decision ----- /home/yeff/public_html/devon/panel/data/panel_canonical_tree.json:6: "ui_reference_mode": "claude_like_workspace_plus_devon_governance" /home/yeff/public_html/devon/panel/data/master_architecture_index.md:109:- ChatGPT memory is NOT a valid source of truth /home/yeff/public_html/devon/panel/data/panel_navigation_spec.json:6: "ui_mode": "claude_like_workspace_devon_governed" /home/yeff/public_html/devon/panel/data/panel_ui_blueprint.json:6: "ui_reference": "claude_like_workspace_plus_devon_governance", /home/yeff/public_html/devon/canon/03_RULES_OF_OPERATION.md:4:O sistema NÃO depende de memória do ChatGPT. /home/yeff/public_html/devon/canon/03_RULES_OF_OPERATION.md:41:- confiar em memória do ChatGPT /home/yeff/public_html/devon/canon/03_RULES_OF_OPERATION.md.bak.align_context_rule_20260409_085318:4:O sistema NÃO depende de memória do ChatGPT. /home/yeff/public_html/devon/canon/03_RULES_OF_OPERATION.md.bak.align_context_rule_20260409_085318:32:- confiar em memória do ChatGPT /home/yeff/public_html/devon/canon/06_DECISIONS.md:18:Impacto: elimina dependência de memória do ChatGPT e reduz drift operacional ===== DECISION MATRIX ===== CURRENT DECISION: - API externa: PROIBIDA. - LLM alvo: Qwen instalado localmente no servidor Devon. - Versão/tamanho/quantização: MISSING até capacidade material do servidor ser avaliada. NEXT DH PATCH SHOULD: - Replace CAS AI cards if they still say "no API/vendor" generically. - Declare Qwen local/self-hosted as the intended LLM family. - Mark specific Qwen version/model size as MISSING until hardware/runtime audit supports a choice. - Keep AI capability classes: reasoning/planning, memory, retrieval, embeddings/vector state, context management, tool execution and orchestration. - Keep DH substrate separate: PHP/JS/JSON/docs/index.php/hub_index.json are documentation/runtime substrate, not the AI core. - Keep JSON mirror blocked until DH is corrected. DO NOT: - Declare OpenAI, GPT, Claude, Gemini or external API. - Declare Qwen version before RAM/VRAM/disk/runtime evidence is evaluated. - Create Technology bucket. - Create JSON mirror before DH cards are corrected. ===== FINAL STATUS ===== PASS: Qwen local capacity audit completed NEXT: replace CAS technology cards with Qwen-local/no-API wording and keep version as MISSING until selected from evidence ===== AUDIT SAVED ===== /home/yeff/public_html/devon/_audit/phase02_cas_qwen_local_capacity_audit_20260429_092427.txt