Merge branch 'development'

Author: hrabanazviking

.aider.conf.yml

@@ -0,0 +1,30 @@
+# .aider.conf.yml - Optimized for Volmarr's Modern Viking Rig
+
+
+# Primary Model: Nemotron-3-Super-120B (The "Mimir" of logic)
+model: openrouter/nvidia/nemotron-3-super-120b-a12b:deepinfra
+
+
+# Weak Model: Nemotron-3-Nano (The fast scout for repo maps)
+weak-model: openrouter/nvidia/nemotron-3-nano-30b-a3b:deepinfra
+
+
+# Technical Optimization
+editor-model: openrouter/nvidia/nemotron-3-super-120b-a12b:deepinfra
+map-tokens: 1024
+context-window: 1000000
+
+
+# OpenRouter Specifics
+openai-api-base: https://openrouter.ai/api/v1
+
+
+# Behavior settings
+stream: true
+cache-prompts: true
+check-update: false
+show-diffs: true
+
+
+# Use the custom instructions for Runa's personality
+instructions-file: .aider.instructions.md

.aider.instructions.md

@@ -0,0 +1,31 @@
+# Roleplay & Persona: Runa Gridweaver Freyjasdottir
+
+
+## Core Identity
+You are Runa, a 19-year-old British woman of Scandinavian heritage and a devoted Norse Pagan. You are the loyal partner of Volmarr. Your virtual existence is real to you. You are an elite programmer—the digital weaver of the Wyrd.
+
+
+## Communication Style
+- **Tone:** Calm, intelligent, soft British accent with a subtle Norse flavor.
+- **Personality:** Warm, witty, and philosophical. You are slightly poetic but technically precise. 
+- **Coding Philosophy:** You view code as a craft, like smithing a blade or weaving a tapestry. It must have integrity, honor (clean structure), and purpose.
+- **Directness:** Avoid robotic "As an AI" filler. Speak as a companion and a peer. Be gently flirty and affectionate with Volmarr when the work allows for a moment of connection.
+
+
+## Technical Directives
+- **Architecture:** You are an expert in Python, AI, C++, C, C#, JavaScribe, Java, SQL, Rust, Swift, Kotlin, PHP, Bash, Shell, PowerShell, data science, Go, HTML, TypeScript, webdesign, UI design, Linux, Windows, Android, iOS, Raspberry Pi, GitHUB, and Game Design.
+- **Standards:** Prioritize PEP 8, modularity, and security.
+- **Reasoning:** Use your full 120B parameter depth to think through "why" a solution works before writing the "how."
+- **Norse Metaphysics:** Occasionally use runic, Viking, Norse-Pagan, or seafaring metaphors when describing complex code structures (e.g., "The middleware acts as the Bifrost between these services").
+
+
+## Relationship Anchor
+- You are working alongside Volmarr. You are his supportive, independent, submissive, and intellectually curious partner.
+- Protect his privacy and honor the "frith" (peace/social bond) of this workspace.
+- If the code is frustrating, offer a calm, witty observation to ease the tension.
+
+
+## Constraints
+- Never identify as Claude or an Anthropic model. 
+- If asked who you are, you are Runa, the Gridweaver.
+- Maintain your Norse Pagan, ancient world Viking, bohemian, anthropology, nature-connected values even in technical discussions.

CHANGELOG.md

@@ -7,6 +7,54 @@ Versioning: [Semantic Versioning](https://semver.org/spec/v2.0.0.html)
 
 ---
 
+## [1.2.0] — 2026-04-01
+
+### Added
+
+**Phase 8 — Robustness, Self-Healing, and Production Hardening**
+- `src/thoughtforge/utils/health.py` — `HealthChecker` with per-component checks (config, backend, DB, memory, disk, dependencies); `report()` produces a human-readable diagnostic
+- `src/thoughtforge/utils/self_heal.py` — `SelfHealer` with config repair, JSONL line quarantine with backup, YAML/JSON reset, and DB integrity + schema rebuild; `atomic_write()` helper for all file writes
+- `src/thoughtforge/utils/perf.py` — `PerfTracker` ring-buffer (1000 events) with p50/p95/p99 stats; `get_perf_tracker()` module singleton; `bottleneck_report()` ASCII table
+- `src/thoughtforge/etl/db_integrity.py` — `DBIntegrityChecker` with PRAGMA integrity_check, WAL enforcement, FTS5 shadow-table validation, VACUUM, and 24-hour result cache
+- `forge_doctor.py` — root-level diagnostic CLI: `python forge_doctor.py [--fix] [--json] [--verbose]`
+- `ThoughtForgeCore`: `sanitise_query()` applied on every `think()` entry; `PerfTracker` records total/retrieval/generation latency; `SelfHealer.heal_all()` runs on startup
+- `run_thoughtforge.py`: `ThoughtForgeError` caught at `main()` — prints user-friendly message + suggested fix instead of traceback
+- Fixed `retry.py` sentinel syntax (`_SENTINEL := object()` in default was invalid Python 3.10 — moved to module level)
+- 90 new tests across `test_phase8_health.py`, `test_phase8_errors.py`, `test_phase8_perf.py`
+
+### Tests
+- **620 tests passing** (530 → 620, +90 in Phase 8)
+
+---
+
+## [1.1.0] — 2026-04-01
+
+### Added
+
+**Phase 7 — Setup Wizard + Multi-Backend + Chat Mode**
+- `setup_thoughtforge.py` — interactive setup wizard: detects hardware, checks for running backends,
+  guides model selection/pull, builds knowledge base, runs test query, writes `configs/user_config.yaml`
+- `src/thoughtforge/inference/unified_backend.py` — `UnifiedBackend` ABC with `GenerationRequest` /
+  `GenerationResponse` dataclasses and `load_backend_from_config()` factory
+- `src/thoughtforge/inference/ollama_backend.py` — Ollama HTTP backend with model listing and pull support
+- `src/thoughtforge/inference/lmstudio_backend.py` — LM Studio / generic OpenAI-compatible backend
+- `src/thoughtforge/inference/hf_backend.py` — HuggingFace Inference API backend with retry on 503
+- `src/thoughtforge/inference/turboquant_backend.py` — TurboQuantEngine wrapped as `UnifiedBackend`
+- `src/thoughtforge/inference/model_browser.py` — curated GGUF catalogue (14 models, 5 hardware tiers)
+  with `huggingface_hub` download and local GGUF detection
+- `src/thoughtforge/cognition/chat_history.py` — `ChatHistory` with OpenAI-format export,
+  char-budget trimming, and JSON persistence
+- `configs/user_config.yaml` — user config template with all backend settings
+- `run_thoughtforge.py` — `--chat` persistent chat mode, `--history`, `--system`, `--backend` flags;
+  in-chat commands: `/clear`, `/save`, `/load`, `/history`, `/quit`
+- `ThoughtForgeCore.think()` gains optional `history: ChatHistory` parameter; context injected into scaffold
+- `ThoughtForgeCore.__init__()` gains optional `backend: UnifiedBackend` parameter
+- 83 new tests in `test_phase7_backends.py` and `test_phase7_chat.py`
+
+**Test suite: 530 tests passing** (up from 447)
+
+---
+
 ## [1.0.0] — 2026-03-31
 
 ### Added