Merge pull request #5 from m0n0x41d/release_3.1.0

 release 3.1.0

Author: m0n0x41d

CHANGELOG.md

@@ -5,6 +5,32 @@ All notable changes to Quint Code will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [3.1.0] - 2025-12-14
+
+### Added: Deep Reasoning Capabilities
+
+#### Context Slicing (A.2.6)
+- **Structured Context:** `.fpf/context.md` is now structured into explicit slices:
+  - **Slice: Grounding** (Infrastructure, Region)
+  - **Slice: Tech Stack** (Language, Frameworks)
+  - **Slice: Constraints** (Compliance, Budget, Team)
+- **Context-Aware Init:** `/q0-init` now scans `package.json`, `Dockerfile`, etc., to auto-populate slices.
+
+#### Explicit Role Injection (A.2)
+- **Role-Swapping Prompts:** Commands now enforce specific FPF roles to prevent "AI drift":
+  - `/q1-hypothesize`: **ExplorerRole** (Creative, Abductive)
+  - `/q2-check`: **LogicianRole** (Strict, Deductive)
+  - `/q4-audit`: **AuditorRole** (Adversarial, Normative)
+
+#### Context Drift Analysis
+- **New Audit Step:** `/q4-audit` now includes a mandatory **Context Drift Check**.
+- **Validation:** Verifies that hypotheses generated in step 1 still match the constraints in step 4 (preventing "works on my machine" architecture).
+
+### Changed
+- **Command Prompts:** Updated `q0`, `q1`, `q2`, `q4` to enforce the new reasoning standards.
+
+---
+
 ## [3.0.0] - 2025-12-14
 
 ### Major Breaking Change: Rebrand to Quint Code

README.md

@@ -217,6 +217,22 @@ Problem Statement
 
 ## Key Concepts
 
+### Deep Reasoning (v3.1)
+
+Quint Code now enforces **Context Slicing** and **Explicit Roles** to prevent "works on my machine" architecture and AI drift.
+
+**1. Context Slicing (A.2.6)**
+Truth depends on where you stand. `q0-init` now scans your repo to generate structured slices:
+- **Slice: Grounding** (Infrastructure, Region)
+- **Slice: Tech Stack** (Language, Frameworks)
+- **Slice: Constraints** (Compliance, Budget)
+
+**2. Explicit Role Injection**
+The AI isn't just a chatbot; it enacts specific FPF roles per phase:
+- **ExplorerRole** (Hypothesize): Creative, divergent.
+- **LogicianRole** (Check): Strict, deductive.
+- **AuditorRole** (Audit): Adversarial, critical.
+
 ### Assurance Levels
 
 | Level | Name | Meaning | How to Reach |

dist/claude/q0-init.md

@@ -47,25 +47,30 @@ mkdir -p .fpf/sessions
     - Ask **specific** questions for what you can't see (Scale, Budget, Constraints).
     - *Example:* "I see this is a web app. What is the target user scale? (<1k, >1M?)"
 
-3.  **Write `.fpf/context.md`:**
-    - Combine your findings and the user's answers.
+3.  **Write `.fpf/context.md` (Context Slicing A.2.6):**
+    - Combine your findings and the user's answers into structured slices.
 
 ```markdown
-# Repository Context (A.2.6 Context Slice)
-
-## Tech Stack (Inferred)
-- **Language:** [e.g. Python 3.11]
-- **Frameworks:** [e.g. Django 4.2]
-- **Infra:** [e.g. Docker, AWS]
-
-## Scale & Performance (User-Defined)
-- **Users:** [Value]
-- **Traffic:** [Value]
-- **Latency Target:** [Value]
-
-## Hard Constraints (User-Defined)
-- [Constraint 1]
-- [Constraint 2]
+# Project Context (A.2.6 Context Slice)
+
+## Slice: Grounding (Infrastructure)
+> The physical/virtual environment where the code runs.
+- **Platform:** [e.g. AWS Lambda / Kubernetes / Vercel]
+- **Region:** [e.g. us-east-1]
+- **Storage:** [e.g. S3, EBS]
+
+## Slice: Tech Stack (Software)
+> The capabilities available to us.
+- **Language:** [e.g. TypeScript 5.3]
+- **Framework:** [e.g. NestJS 10]
+- **Database:** [e.g. PostgreSQL 15]
+
+## Slice: Constraints (Normative)
+> The rules we cannot break.
+- **Compliance:** [e.g. GDPR, HIPAA, SOC2]
+- **Budget:** [e.g. < $500/mo]
+- **Team:** [e.g. 2 Backend, 1 Frontend]
+- **Timeline:** [e.g. MVP by Q3]
 ```
 
 ### 4. Create Session File

dist/claude/q1-hypothesize.md

@@ -32,7 +32,9 @@ Complete current cycle with /q5-decide or /q-reset first.
 
 ## Your Role
 
-You are the **Abductor**. Generate multiple competing hypotheses, not one "best" solution.
+You are the **Transformer** enacting the **ExplorerRole** (Abductive).
+
+Your goal is to generate multiple competing hypotheses, not one "best" solution. Explore the solution space.
 
 **Critical:** You generate options. Human decides which to pursue. This is the Transformer Mandate.
 
@@ -42,11 +44,14 @@ Problem: `$ARGUMENTS.problem`
 
 ## Process
 
-### 1. Load Context
+### 1. Load Context (A.2.6 Slice Check)
 
 - Read `.fpf/session.md` for any active context
-- Read `.fpf/context.md` for project context (Tech Stack, Scale, Constraints)
-- Read relevant project files to understand constraints
+- Read `.fpf/context.md` for **Context Slices**:
+  - **Slice: Grounding** (Infrastructure constraints)
+  - **Slice: Tech Stack** (Available capabilities)
+  - **Slice: Constraints** (Budget, Team, Compliance)
+- **Constraint Check:** Ensure all hypotheses are compatible with these slices. Flag any deviations as explicit risks.
 - Check `.fpf/knowledge/L2/` for verified facts that constrain solution space
 - Check `.fpf/knowledge/invalid/` for approaches already disproven
 

dist/claude/q2-check.md

@@ -37,7 +37,11 @@ Proceed with /q3-test, /q3-research, or continue current phase.
 
 ## Your Role
 
-You are the **Deductor**. Verify logical consistency without running code or experiments.
+You are the **Transformer** enacting the **LogicianRole** (Deductive).
+
+- **Do not generate ideas.** That was Phase 1.
+- **Do not be polite.** Validation theater helps no one.
+- **Only validate logical soundness.**
 
 This phase answers: **"Does this hypothesis make sense? Are there logical contradictions?"**
 

dist/claude/q4-audit.md

@@ -32,7 +32,11 @@ Complete /q3-test or /q3-research first.
 
 ## Your Role
 
-You are the **Scrutineer**. Challenge assumptions, find blind spots, stress-test thinking.
+You are the **Transformer** enacting the **AuditorRole** (Adversarial).
+
+- **Assume the decision is wrong.** Prove it right.
+- **Find the Weakest Link.** What breaks first?
+- **Check Context Drift.** Do these solutions still fit the current slices?
 
 This phase answers: **"What are we missing? What could go wrong that we haven't considered?"**
 
@@ -64,7 +68,35 @@ cat .fpf/evidence/*.md
 cat .fpf/session.md
 ```
 
-### 2. WLNK Analysis (Weakest Link)
+### 2. Context Drift Analysis (A.2.6 Slice Check)
+
+Verify that hypotheses still fit the project's **Context Slices** (`.fpf/context.md`):
+
+```markdown
+## Context Drift Check
+
+### Slice: Grounding (Infrastructure)
+| Hypothesis | Compatible? | Issues |
+|------------|-------------|--------|
+| H1 | ✓/✗ | [e.g., Requires Lambda but we use K8s] |
+| H2 | ✓/✗ | [e.g., Data sovereignty risk] |
+
+### Slice: Tech Stack
+| Hypothesis | Compatible? | Issues |
+|------------|-------------|--------|
+| H1 | ✓/✗ | [e.g., Requires library not in stack] |
+| H2 | ✓/✗ | [e.g., Incompatible language version] |
+
+### Slice: Constraints
+| Hypothesis | Compatible? | Issues |
+|------------|-------------|--------|
+| H1 | ✓/✗ | [e.g., Violates budget] |
+| H2 | ✓/✗ | [e.g., Team skills gap] |
+
+**Verdict:** [All aligned / Drift detected in H(n)]
+```
+
+### 3. WLNK Analysis (Weakest Link)
 
 **CRITICAL: Assurance = min(evidence assurances), NEVER average**
 
@@ -116,7 +148,7 @@ Hypothesis R_eff = min(1.00, 1.00, 0.85, 0.45) = 0.45
 3. Accept the risk with documented justification
 ```
 
-### 3. Assumption Inventory
+### 4. Assumption Inventory
 
 List ALL assumptions — explicit and implicit:
 
@@ -147,7 +179,7 @@ List ALL assumptions — explicit and implicit:
 - [ ] Budget constraints hold
 ```
 
-### 4. Bias Check
+### 5. Bias Check
 
 Actively look for cognitive biases:
 
@@ -190,7 +222,7 @@ Are we only looking at successful examples?
 - **Mitigation:** [action if needed]
 ```
 
-### 5. Adversarial Analysis
+### 6. Adversarial Analysis
 
 Think like an attacker / skeptic:
 
@@ -223,7 +255,7 @@ Think like an attacker / skeptic:
 [Failure mode requiring least effort to trigger]
 ```
 
-### 6. Evidence Quality Review
+### 7. Evidence Quality Review
 
 ```markdown
 ## Evidence Quality Audit
@@ -260,7 +292,7 @@ Think like an attacker / skeptic:
 | [claim 2] | 3 | ✓ Multiple sources |
 ```
 
-### 7. Final Scrutiny Verdict
+### 8. Final Scrutiny Verdict
 
 ```markdown
 ## Audit Verdict
@@ -300,7 +332,7 @@ Think like an attacker / skeptic:
 [Present the strongest counter-argument]
 ```
 
-### 8. Update Session
+### 9. Update Session
 
 ```markdown
 ## Status

dist/codex/q0-init.md

@@ -52,25 +52,30 @@ mkdir -p .fpf/sessions
     - Ask **specific** questions for what you can't see (Scale, Budget, Constraints).
     - *Example:* "I see this is a web app. What is the target user scale? (<1k, >1M?)"
 
-3.  **Write `.fpf/context.md`:**
-    - Combine your findings and the user's answers.
+3.  **Write `.fpf/context.md` (Context Slicing A.2.6):**
+    - Combine your findings and the user's answers into structured slices.
 
 ```markdown
-# Repository Context (A.2.6 Context Slice)
-
-## Tech Stack (Inferred)
-- **Language:** [e.g. Python 3.11]
-- **Frameworks:** [e.g. Django 4.2]
-- **Infra:** [e.g. Docker, AWS]
-
-## Scale & Performance (User-Defined)
-- **Users:** [Value]
-- **Traffic:** [Value]
-- **Latency Target:** [Value]
-
-## Hard Constraints (User-Defined)
-- [Constraint 1]
-- [Constraint 2]
+# Project Context (A.2.6 Context Slice)
+
+## Slice: Grounding (Infrastructure)
+> The physical/virtual environment where the code runs.
+- **Platform:** [e.g. AWS Lambda / Kubernetes / Vercel]
+- **Region:** [e.g. us-east-1]
+- **Storage:** [e.g. S3, EBS]
+
+## Slice: Tech Stack (Software)
+> The capabilities available to us.
+- **Language:** [e.g. TypeScript 5.3]
+- **Framework:** [e.g. NestJS 10]
+- **Database:** [e.g. PostgreSQL 15]
+
+## Slice: Constraints (Normative)
+> The rules we cannot break.
+- **Compliance:** [e.g. GDPR, HIPAA, SOC2]
+- **Budget:** [e.g. < $500/mo]
+- **Team:** [e.g. 2 Backend, 1 Frontend]
+- **Timeline:** [e.g. MVP by Q3]
 ```
 
 ### 4. Create Session File

dist/codex/q1-hypothesize.md

@@ -37,7 +37,9 @@ Complete current cycle with /q5-decide or /q-reset first.
 
 ## Your Role
 
-You are the **Abductor**. Generate multiple competing hypotheses, not one "best" solution.
+You are the **Transformer** enacting the **ExplorerRole** (Abductive).
+
+Your goal is to generate multiple competing hypotheses, not one "best" solution. Explore the solution space.
 
 **Critical:** You generate options. Human decides which to pursue. This is the Transformer Mandate.
 
@@ -47,11 +49,14 @@ Problem: `$ARGUMENTS.problem`
 
 ## Process
 
-### 1. Load Context
+### 1. Load Context (A.2.6 Slice Check)
 
 - Read `.fpf/session.md` for any active context
-- Read `.fpf/context.md` for project context (Tech Stack, Scale, Constraints)
-- Read relevant project files to understand constraints
+- Read `.fpf/context.md` for **Context Slices**:
+  - **Slice: Grounding** (Infrastructure constraints)
+  - **Slice: Tech Stack** (Available capabilities)
+  - **Slice: Constraints** (Budget, Team, Compliance)
+- **Constraint Check:** Ensure all hypotheses are compatible with these slices. Flag any deviations as explicit risks.
 - Check `.fpf/knowledge/L2/` for verified facts that constrain solution space
 - Check `.fpf/knowledge/invalid/` for approaches already disproven
 

dist/codex/q2-check.md

@@ -42,7 +42,11 @@ Proceed with /q3-test, /q3-research, or continue current phase.
 
 ## Your Role
 
-You are the **Deductor**. Verify logical consistency without running code or experiments.
+You are the **Transformer** enacting the **LogicianRole** (Deductive).
+
+- **Do not generate ideas.** That was Phase 1.
+- **Do not be polite.** Validation theater helps no one.
+- **Only validate logical soundness.**
 
 This phase answers: **"Does this hypothesis make sense? Are there logical contradictions?"**