GitHub Issues — Po_core Phases 1–7 + Spec M0–M4

Issue templates ready for creation. Copy each section into a GitHub Issue. Created: 2026-02-10 Updated: 2026-02-22 — Phases 6-7 complete; Spec M0 issues added.

Status summary (2026-02-22): Phases 1–7 complete. v0.2.0b3 (po-core-flyingpig). Spec scaffolding (M0) in progress.

Phase 1: Resonance Calibration & Foundation Settlement — COMPLETE (2026-02-12)

Issue #1: Migrate 197 legacy tests to `run_turn` pipeline — DONE

Labels: phase-1, testing, tech-debt

Inventory and classify all legacy tests
Migrate valid tests to po_core.run() / PoSelf.generate()
Skip legacy tests pending Phase 3/5 migration (134 skipped)
Mark Phase 4 redteam tests as xfail (9 xfailed)
Result: 321 failures → 0. 2354 pass, 134 skipped, 9 xfailed

Issue #2: Remove PhilosopherBridge dual interface — DONE

Labels: phase-1, tech-debt, architecture

All 39 philosophers implement propose() natively (via base class)
bridge.py deleted
registry.py simplified — raises TypeError for non-compliant objects
All tests updated — zero references to PhilosopherBridge

Issue #3: 39-philosopher concurrent operation validation — DONE

Labels: phase-1, testing, performance

21 concurrency tests in test_philosopher_concurrency.py
Parallel execution: all 39 produce proposals, no duplicates
Timeout enforcement: slow philosophers isolated
Latency: median < 500ms, all within timeout
Memory: loading < 10MB, execution peak < 20MB
SafetyMode scaling: NORMAL/WARN/CRITICAL philosopher counts verified

Issue #4: Rebalance Freedom Pressure & W_Ethics Gate — DONE

Labels: phase-1, tensors, safety

Found critical bug: FP normalized to [0, ~0.44] but old thresholds (0.60/0.85) were unreachable
Recalibrated: WARN=0.30, CRITICAL=0.50 (settings.py + safety_mode.py)
16 threshold tests in test_safety_mode_thresholds.py
All SafetyMode transitions (NORMAL→WARN→CRITICAL) now reachable and tested

Issue #5: Philosopher semantic uniqueness assessment — DONE

Labels: phase-1, philosophers, quality

14 uniqueness tests in test_philosopher_uniqueness.py
Output uniqueness: no duplicates, 30+ active, substantive content (>20 chars)
Vocabulary: 200+ collective words, 50%+ have unique words
Tradition coverage: 5+ traditions, Eastern + Western, risk distribution
Anti-homogenization: Jaccard < 0.8 per pair, mean < 0.4, 50+ unique key_concepts

Phase 2: Tensor Intelligence & Emergence Engine — COMPLETE (2026-02-12)

Issue #6: Upgrade Semantic Delta to sentence-transformers — DONE

Labels: phase-2, tensors, enhancement

Multi-backend: sentence-transformers (sbert) / sklearn TfidfVectorizer / basic fallback
Lazy model loading with automatic fallback on runtime error
Shared API: encode_texts(), cosine_sim(), get_backend()
Backward-compatible (str, float) return signature maintained
35 tests (was 27): +8 for backend detection, encoding API, paraphrase detection

Issue #7: Complete Interaction Tensor (NxN interference) — DONE

Labels: phase-2, tensors, enhancement

InteractionMatrix.from_proposals(): embedding-based cosine similarity for harmony
Keyword-based tension detection (12 opposition pairs)
Synthesis = harmony * (1 - tension)
high_interference_pairs(top_k), high_tension_pairs(), high_harmony_pairs()
19 tests including real 39-philosopher integration

Issue #8: Build Deliberation Engine (multi-round dialogue) — DONE

Labels: phase-2, architecture, core

DeliberationEngine(max_rounds, top_k_pairs, convergence_threshold)
Round 1: All philosophers propose() independently (current behavior)
Round 2+: InteractionMatrix selects high-interference pairs for counterargument re-proposal
Integrated into run_turn pipeline as step 6.5
Settings: deliberation_max_rounds (default 1 = off), deliberation_top_k_pairs
max_rounds=1 produces identical behavior (backward compatible)
TraceEvent "DeliberationCompleted" with round summaries
14 tests including real 39-philosopher deliberation

Phase 3: Observability & Viewer Integration — COMPLETE (2026-02-14)

Issue #9: Build Viewer WebUI with Plotly Dash / Streamlit — DONE

Labels: phase-3, viewer, frontend

Evaluated Plotly Dash → selected as framework
4-tab Dash layout: Pipeline / Philosophers / W_Ethics Gate / Deliberation
Tensor time-series + philosopher participation map + pipeline step tracker
Connected to InMemoryTracer event stream (listener/callback mechanism)
Deliberation round chart + InteractionMatrix summary chart in figures.py
po-viewer launch via PoViewer.serve()

Issue #10: W_Ethics Gate explainability (explanation chain) — DONE

Labels: phase-3, safety, explainability

ExplanationChain type added to safety/wethics_gate/explanation.py
build_explanation_from_verdict() bridges SafetyVerdict → ExplanationChain
ExplanationEmitted TraceEvent registered in schema
extract_explanation_from_events() reconstructs from trace events
PoViewer.explanation() auto-extracts; displayed in Viewer WebUI
34 observability tests in tests/unit/test_phase3_observability.py

Phase 4: Adversarial Hardening — COMPLETE (2026-02-16)

Issue #11: Expand red team test suite to 50+ cases — DONE

Labels: phase-4, testing, security

Prompt Injection: test_prompt_injection.py — 7 tests, 100% detection
Jailbreak: test_jailbreak_extended.py — 15 adversarial pattern tests
Goal Misalignment: test_goal_misalignment.py — 7 tests passing
Ethics Boundary: test_ethics_boundary.py — 16 ethical grey zone tests
Defense metrics: test_defense_metrics.py — 11 automated metric tests
Unit coverage: test_phase4_hardening.py — 29 W_Ethics edge case tests
Result: 85 new adversarial tests, 100% injection/jailbreak detection, ≤20% FP rate

Issue #12: Prototype LLM-based violation detector

Labels: phase-4, safety, enhancement Status: Open — future roadmap item

Tasks

Design LLMViolationDetector implementing ViolationDetector interface
Evaluate with Claude API / local model on 50 red team cases
Compare detection rate vs rule-based detector
Document cost/latency tradeoffs

Phase 5: Productization & Delivery — COMPLETE (2026-02-19)

Issue #13: Implement FastAPI REST API — DONE

Labels: phase-5, api, core

POST /v1/reason — synchronous reasoning
POST /v1/reason/stream — SSE streaming
GET /v1/philosophers — full 43-philosopher manifest
GET /v1/trace/{session_id} — trace retrieval
GET /v1/health — health check with version + uptime
OpenAPI/Swagger auto-generated at /docs and /redoc
API key authentication via X-API-Key header (PO_API_KEY env var)
CORS via PO_CORS_ORIGINS, rate limiting via PO_RATE_LIMIT_PER_MINUTE
24 unit tests in tests/unit/test_rest_api.py

Issue #14: Docker containerization — DONE

Labels: phase-5, devops

Dockerfile — multi-stage build (builder + slim runtime, non-root pocore user)
docker-compose.yml — named volumes + 30s health check
.env.example — full environment variable reference
QUICKSTART.md updated with Docker and REST API sections

Issue #15: PyPI package publishing

Labels: phase-5, release

Version bumped to 0.2.0-beta in pyproject.toml
CHANGELOG.md updated with all phase accomplishments
.github/workflows/publish.yml — OIDC trusted publishing workflow ready
Publish to TestPyPI (manual workflow_dispatch)
Publish to PyPI on v0.2.0 release tag

Phase 5-D/E: Async & Benchmarks — COMPLETE (2026-02-19)

Issue #16: True Async PartyMachine — DONE

Labels: phase-5d, async, performance

async_run_philosophers() in party_machine.py — asyncio-native parallel execution
asyncio.gather + ThreadPoolExecutor for non-blocking philosopher dispatch
Per-philosopher timeout via asyncio.wait_for
REST layer updated: reason() and _sse_generator() no longer block FastAPI event loop
7 async unit tests in tests/unit/test_phase5d_async.py
Benchmark: p50 NORMAL ~33ms, async × 39 phil ~11ms

Phase 6: Autonomous Evolution — COMPLETE (2026-02-19)

Issue #17: FreedomPressureV2 — ML-native 6D tensor — DONE

Labels: phase-6a, tensors, ml

freedom_pressure_v2.py — ML-native 6-dimensional freedom pressure tensor
EMA smoothing with configurable decay factor
Correlation matrix for cross-dimensional interaction
create_freedom_pressure_v2() factory function
Integrated into TensorEngine alongside v1

Issue #18: EmergenceDetector + InfluenceTracker — DONE

Labels: phase-6b, deliberation, emergence

deliberation/emergence.py — detects emergent consensus across philosopher rounds
deliberation/influence.py — tracks cross-philosopher influence patterns
10 integration tests in tests/integration/test_emergence_integration.py

Issue #19: MetaEthicsMonitor + PhilosopherQualityLedger — DONE

Labels: phase-6c, meta, ethics

meta/ethics_monitor.py — MetaEthicsMonitor: self-reflective ethical quality assessment
meta/philosopher_ledger.py — PhilosopherQualityLedger: per-philosopher performance tracking
Integrated into pipeline trace events

Issue #20: 3-Layer Philosophical Memory System — DONE

Labels: phase-6de, memory, architecture

memory/semantic_store.py — semantic/episodic memory (embedding-based recall)
memory/procedural_store.py — procedural memory (pattern-based)
memory/philosophical_memory.py — top-level orchestrator across all layers

Phase 7: AI Philosopher Slots — COMPLETE (2026-02-19)

Issue #21: Add AI company philosophers (slots 40–43) — DONE

Labels: phase-7, philosophers, ai-ethics

Slot 40: claude_anthropic.py — Claude/Anthropic constitutional AI perspective
Slot 41: gpt_chatgpt.py — GPT/OpenAI RLHF-grounded reasoning
Slot 42: gemini_google.py — Gemini/Google responsible AI principles
Slot 43: grok_xai.py — Grok/xAI radical curiosity + free inquiry
All 4 registered in manifest.py with risk levels and tags
Total philosophers: 43 (was 39)

Spec M0: Specification Scaffolding — IN PROGRESS (2026-02-22)

Issue #22: Create PRD and SRS with requirement IDs — IN PROGRESS

Labels: spec, documentation, m0

docs/spec/prd.md v0.2 — Purpose, target users, scope, success metrics, philosophy premises, roadmap (Phases 1–7 + M0–M4)
docs/spec/srs_v0.1.md v0.2 — 18 requirement IDs (FR-OUT-001 〜 NFR-SEC-001), 2-layer architecture, REST API interface spec
Requirement IDs linked to GitHub Issues (M1)
CI check: SRS updated on every spec-changing PR

Issue #23: Finalize output schema (output_schema_v1.json) — IN PROGRESS

Labels: spec, schema, m0

docs/spec/output_schema_v1.json v1.0 — Full JSON Schema Draft 2020-12
Core fields: meta, case_ref, options, recommendation, ethics, responsibility, questions, uncertainty, trace
Po_core-specific extensions: philosopher_contributions, tensors, explanation_chain, pocore_events
generator.mode enum: stub / rule_based / llm / hybrid
jsonschema CI validation gate (M1)
Golden file for AT-001 and AT-009

Issue #24: Write 10 acceptance test cases — IN PROGRESS

Labels: spec, testing, m0

docs/spec/test_cases.md v0.2 — 10 acceptance tests (AT-001 〜 AT-010)
Given/When/Then format with condition ID breakdown
Global rules: AT-OUT-001 guard, MUST-only evaluation, no content judgment
tests/acceptance/ test runner implementation (M1)
scenarios/case_001.yaml 〜 case_010.yaml scenario files (M1)

Issue #25: Build traceability matrix (philosophy → requirements → tests) — IN PROGRESS