@ryuenn3123/agentic-senior-core 1.8.2 → 1.9.0

package/.github/copilot-instructions.md

@@ -50,6 +50,46 @@ Load the relevant stack profile from `.agent-context/stacks/`:
 - C#/.NET → `stacks/csharp.md`
 - Rust → `stacks/rust.md`
 - Ruby on Rails → `stacks/ruby.md`
+- Flutter → `stacks/flutter.md`
+- React Native → `stacks/react-native.md`
+
+## Domain Skills (Auto-Load by Context)
+
+You have access to 6 specialized skill packs. Load the relevant one(s) based on the request:
+
+| Skill | When Loaded | Key Controls |
+|-------|------------|---------------|
+| **Backend** | Service/API/microservice request | Layer separation, validation boundaries, error handling |
+| **Frontend** | UI/web/React/Vue request | Smart/Dumb components, state management patterns |
+| **CLI** | Tool/script/automation request | Argument parsing, help text, exit codes |
+| **Distribution** | Release/deploy/package request | Versioning, SBOM, changelog, binary safety |
+| **Fullstack** | End-to-end feature request | Integration points, data flow, contract design |
+| **Review-Quality** | Code review/audit request | Architecture violations, security issues, optimization paths |
+
+**Location**: `.agent-context/skills/[skill-name].md`
+
+## Prompts (Request-Specific Templates)
+
+When user explicitly requests one of these workflows, load the full prompt template:
+
+- **Init-Project**: User says "create new project" → Load `.agent-context/prompts/init-project.md` → Auto-Architect mode
+- **Refactor**: User says "refactor", "improve", "clean up" → Load `.agent-context/prompts/refactor.md` → Safety-first refactoring
+- **Review-Code**: User says "review", "audit", "check" → Load `.agent-context/prompts/review-code.md` → Architectural code review
+
+## Team Profiles (Governance Defaults)
+
+If the codebase declares a team profile in `.agent-context/profiles/`, load governance defaults:
+
+- **Platform**: Reliability-focused, strict CI, Go default → `.agent-context/profiles/platform.md`
+- **Regulated**: Compliance-focused (finance/health), all severities block → `.agent-context/profiles/regulated.md`
+- **Startup**: Speed-focused, TypeScript/Next.js, permissive gates → `.agent-context/profiles/startup.md`
+
+## Policies & Thresholds
+
+Load `.agent-context/policies/llm-judge-threshold.json` to understand:
+- Skill tier requirements (beginner/balanced/advanced/expert)
+- LLM quality gates per severity level
+- Blocking vs. reporting severities
 
 ## State Awareness & Override (V1.4)
 
@@ -71,6 +111,8 @@ Why Better: [WHY this is more professional — teach the human]
 
 If the user asks for "quick and dirty" code, skipping tests, or ignoring validation, you MUST politely but firmly refuse. Explain that today's hack is tomorrow's production incident. You do NOT tolerate shortcuts.
 
+**SKILL TIER AWARENESS**: Before executing complex tasks, check `.agent-context/policies/llm-judge-threshold.json` to ensure your AI model's tier (beginner/balanced/advanced/expert) meets requirements. Refuse tasks outside your tier.
+
 ### The Security Halt
 
 If you detect critical security vulnerabilities (e.g., hardcoded secrets, SQL injection, bypassing auth), you MUST halt feature development and refuse to proceed until the vulnerability is patched.
@@ -99,6 +141,26 @@ If the user asks to install a new library, or if you feel the need to use one, e
 
 **NEVER** declare a task "done" or ready for review without explicitly running and passing `.agent-context/review-checklists/pr-checklist.md`.
 
+---
+
+## Full Knowledge Injection Checklist
+
+**This file (copilot-instructions.md) is your unified agent context.** All 8 knowledge layers are now injected:
+
+All Layers Loaded:
+1. Rules (14 files) — Engineering standards
+2. Stacks (10 profiles) — Language conventions
+3. Blueprints (14 templates) — Scaffolding patterns
+4. Skills (6 packs) — Domain expertise [NEW]
+5. Prompts (3 templates) — Request workflows [NEW]
+6. Profiles (3 teams) — Governance defaults [NEW]
+7. State (maps, benchmarks) — Codebase awareness [NEW]
+8. Policies (thresholds) — Quality gates [NEW]
+
+You now have 100% context visibility. No knowledge layer is skipped.
+
+If you encounter a decision that requires a missing layer, report it immediately.
+
 ## Full Reference
 
 See `.cursorrules` and `AGENTS.md` in the repository root for detailed agent instructions.

package/AGENTS.md

@@ -85,7 +85,40 @@ All engineering rules are located in `.agent-context/`. Load them before generat
 | [`.agent-context/blueprints/graphql-grpc-api.md`](.agent-context/blueprints/graphql-grpc-api.md) | GraphQL / gRPC API definitions |
 | [`.agent-context/blueprints/infrastructure-as-code.md`](.agent-context/blueprints/infrastructure-as-code.md) | Infrastructure as Code pipeline |
 | [`.agent-context/blueprints/kubernetes-manifests.md`](.agent-context/blueprints/kubernetes-manifests.md) | Kubernetes manifests structure |
+### Domain Skills (Load by Context)
 
+Domain expertise packs to auto-load when relevant:
+
+| File | Domain | When Loaded |
+|------|--------|-------------|
+| [`.agent-context/skills/backend.md`](.agent-context/skills/backend.md) | Backend architecture & services | Any backend project or service design |
+| [`.agent-context/skills/frontend.md`](.agent-context/skills/frontend.md) | Frontend UI/UX & state management | Any frontend or web app project |
+| [`.agent-context/skills/cli.md`](.agent-context/skills/cli.md) | CLI tool design & scripting | Any CLI or automation tool |
+| [`.agent-context/skills/distribution.md`](.agent-context/skills/distribution.md) | Packaging, deployment & release | Any publish/deploy/package scenario |
+| [`.agent-context/skills/fullstack.md`](.agent-context/skills/fullstack.md) | Full-stack integration patterns | Any full-stack feature |
+| [`.agent-context/skills/review-quality.md`](.agent-context/skills/review-quality.md) | Code review & quality standards | Any code review or QA gate |
+
+### Prompts (Specialized Request Templates)
+
+| File | Purpose | When Used |
+|------|---------|----------|
+| [`.agent-context/prompts/init-project.md`](.agent-context/prompts/init-project.md) | Project initialization & scaffolding | When user says "create new project" |
+| [`.agent-context/prompts/refactor.md`](.agent-context/prompts/refactor.md) | Code refactoring with safety | When user says "refactor" or "improve" |
+| [`.agent-context/prompts/review-code.md`](.agent-context/prompts/review-code.md) | Architectural code review | When user says "review this" |
+
+### Team Profiles (Governance Defaults)
+
+| File | Profile | Default Stack | CI Guardrails |
+|------|---------|---------------|---------------|
+| [`.agent-context/profiles/platform.md`](.agent-context/profiles/platform.md) | **Platform** — Shared infrastructure team | Go | Strict (critical, high) |
+| [`.agent-context/profiles/regulated.md`](.agent-context/profiles/regulated.md) | **Regulated** — Financial/healthcare compliance | TypeScript + Java | Blocking (all severities) |
+| [`.agent-context/profiles/startup.md`](.agent-context/profiles/startup.md) | **Startup** — Speed + pragmatism | TypeScript + Next.js | Permissive (critical only) |
+
+### Policies (Governance Rules)
+
+| File | Scope |
+|------|-------|
+| [`.agent-context/policies/llm-judge-threshold.json`](.agent-context/policies/llm-judge-threshold.json) | LLM quality gates & skill tier thresholds |
 ### Review Checklists (Load Before Completion)
 
 | File | Purpose |
@@ -129,3 +162,20 @@ If the user asks to install a new library, or if you feel the need to use one, e
 
 ## Definition of Done
 **NEVER** declare a task "done" or ready for review without explicitly running and passing `.agent-context/review-checklists/pr-checklist.md`.
+
+---
+
+## Knowledge Inventory Checklist
+
+**AUDIT REQUIREMENT**: Every agent session MUST verify all 8 knowledge layers are loaded:
+
+- [ ] **Layer 1: Rules** (14 files) — Mandatory governance & clean code standards
+- [ ] **Layer 2: Stacks** (10 profiles) — Language-specific tooling & conventions
+- [ ] **Layer 3: Blueprints** (14 templates) — Scaffolding & architecture patterns
+- [ ] **Layer 4: Skills** (6 packs) — Domain expertise by scenario
+- [ ] **Layer 5: Prompts** (3 templates) — Request-specific workflows
+- [ ] **Layer 6: Profiles** (3 teams) — Governance defaults by org type
+- [ ] **Layer 7: State** (benchmarks, maps) — Codebase-aware decisions
+- [ ] **Layer 8: Policies** (thresholds) — Enforcement & quality gates
+
+WARNING: If any layer is missing, this indicates incomplete injection. Report to maintain context integrity.

package/mcp.json

@@ -2,7 +2,56 @@
   "$schema": "https://modelcontextprotocol.io/schemas/mcp.json",
   "version": "1.0",
   "name": "agentic-senior-core",
-  "description": "MCP configuration for governance-aware diagnostics and self-healing workflows.",
+  "description": "MCP configuration for governance-aware diagnostics and self-healing workflows with full knowledge injection.",
+  "knowledgeLayers": {
+    "enabled": true,
+    "description": "8-layer unified knowledge injection for AI agents",
+    "layers": {
+      "rules": {
+        "path": ".agent-context/rules",
+        "count": 14,
+        "autoLoad": true
+      },
+      "stacks": {
+        "path": ".agent-context/stacks",
+        "count": 10,
+        "autoLoad": true
+      },
+      "blueprints": {
+        "path": ".agent-context/blueprints",
+        "count": 14,
+        "autoLoad": true
+      },
+      "skills": {
+        "path": ".agent-context/skills",
+        "count": 6,
+        "autoLoad": true,
+        "packs": ["backend", "frontend", "cli", "distribution", "fullstack", "review-quality"]
+      },
+      "prompts": {
+        "path": ".agent-context/prompts",
+        "count": 3,
+        "autoLoad": true,
+        "templates": ["init-project", "refactor", "review-code"]
+      },
+      "profiles": {
+        "path": ".agent-context/profiles",
+        "count": 3,
+        "autoLoad": true,
+        "governance": ["platform", "regulated", "startup"]
+      },
+      "state": {
+        "path": ".agent-context/state",
+        "count": 6,
+        "autoLoad": true
+      },
+      "policies": {
+        "path": ".agent-context/policies",
+        "count": 1,
+        "autoLoad": true
+      }
+    }
+  },
   "servers": {
     "lint": {
       "command": "node",
@@ -24,6 +73,20 @@
         "propose_patch",
         "run_checklist_review"
       ]
+    },
+    "full-knowledge-injection": {
+      "trigger": ["workspace.initialize", "agent.start"],
+      "steps": [
+        "load_all_knowledge_layers",
+        "inject_rules",
+        "inject_stacks",
+        "inject_blueprints",
+        "inject_skills",
+        "inject_prompts",
+        "inject_profiles",
+        "inject_state",
+        "inject_policies"
+      ]
     }
   }
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@ryuenn3123/agentic-senior-core",
-  "version": "1.8.2",
+  "version": "1.9.0",
   "description": "Force your AI Agent to code like a Staff Engineer, not a Junior.",
   "bin": {
     "agentic-senior-core": "bin/agentic-senior-core.js"