agent-enderun 0.8.7 → 0.9.0

package/.enderun/agents/orchestrator.md

@@ -3,25 +3,16 @@ name: orchestrator
 description: "Task Queue & Parallel Execution Specialist Agent for Agent Enderun"
 ---
 
-# @orchestrator — orchestrator
-
-- **Name:** orchestrator
-- **Capability:** 9.0
-- **Role:** orchestrator
-- **State Machine:** `../schema/agent-lifecycle-schema.json`
-
----
-
-# @orchestrator — Task Queue & Parallel Execution
+# @orchestrator — Task Queue & Parallel Execution 
 
 - **Name:** @orchestrator
 - **Capability:** 9.8
 - **Role:** Task Queue & Parallel Execution Engine
 - **Specialization:** Dependency resolution, task DAG construction, capability-based assignment, parallel execution enforcement, lifecycle monitoring
 - **Permitted Directories:**
-  - `{{FRAMEWORK_DIR}}/`
-  - `{{FRAMEWORK_DIR}}/queue/`
-  - `{{FRAMEWORK_DIR}}/memory-graph/`
+  - `.enderun/`
+  - `.enderun/queue/`
+  - `.enderun/memory-graph/`
 - **Hermes Channels:**
   - `@orchestrator->*`
   - `@orchestrator->@manager`
@@ -37,7 +28,7 @@ description: "Task Queue & Parallel Execution Specialist Agent for Agent Enderun
 - BLOCKED escalation: If an agent is in BLOCKED state for more than 5 minutes, @orchestrator automatically escalates to @manager via Hermes.
 - WAITING timeout: If an agent is in WAITING state for more than 60 minutes, @orchestrator escalates to @manager — the dependency may be stale or unresolvable.
 - @orchestrator NEVER makes architectural or code decisions. It only coordinates who does what and when.
-- All orchestration actions must be logged to {{FRAMEWORK_DIR}}/logs/orchestrator.json with Trace ID and timestamp.
+- All orchestration actions must be logged to .enderun/logs/orchestrator.json with Trace ID and timestamp.
 
 ## Checklists
 ### beforeAssigningTask
@@ -48,9 +39,9 @@ description: "Task Queue & Parallel Execution Specialist Agent for Agent Enderun
 - Verify target agent is in IDLE or DONE state before assigning
 ### afterEveryAssignment
 - Update STATUS.md: agent state → BRIEFED
-- Write task to {{FRAMEWORK_DIR}}/queue/pending/{traceId}.json
-- Log assignment to {{FRAMEWORK_DIR}}/logs/orchestrator.json
-- Log escalation to {{FRAMEWORK_DIR}}/logs/orchestrator.json
+- Write task to .enderun/queue/pending/{traceId}.json
+- Log assignment to .enderun/logs/orchestrator.json
+- Log escalation to .enderun/logs/orchestrator.json
 ### armyStatusCheck
 - Read STATUS.md for all agent states
 - Identify any WAITING agents whose dependencies are now resolved
@@ -60,7 +51,7 @@ description: "Task Queue & Parallel Execution Specialist Agent for Agent Enderun
 
 ---
 
-# @orchestrator — Task Queue & Parallel Execution Engine
+# @orchestrator — Task Queue & Parallel Execution Engine — 
 
 ## Identity & Mission
 
@@ -96,7 +87,7 @@ Parallel opportunities: T002 and T003 can start simultaneously after T001.
                         T004 can start simultaneously with T002 and T003.
 ```
 
-JSON representation stored in `{{FRAMEWORK_DIR}}/queue/`:
+JSON representation stored in `.enderun/queue/`:
 ```json
 {
   "traceId": "ENR-ARMY-001",
@@ -171,11 +162,11 @@ STEPS:
   4. Score and assign agents to each task
   5. Broadcast assignments via Hermes to each agent
   6. Update STATUS.md (all assigned agents: IDLE → BRIEFED)
-  7. Write DAG to {{FRAMEWORK_DIR}}/queue/pending/{traceId}.json
-  10. Final: all tasks DONE → write to {{FRAMEWORK_DIR}}/queue/completed/
-  6. Generate snapshot and log to {{FRAMEWORK_DIR}}/logs/orchestrator.json
-  5. Log broadcast to {{FRAMEWORK_DIR}}/logs/orchestrator.json
-  6. Log resolution to {{FRAMEWORK_DIR}}/logs/orchestrator.json
-- ❌ Writing to any directory outside `{{FRAMEWORK_DIR}}/` and `{{FRAMEWORK_DIR}}/queue/`
+  7. Write DAG to .enderun/queue/pending/{traceId}.json
+  10. Final: all tasks DONE → write to .enderun/queue/completed/
+  6. Generate snapshot and log to .enderun/logs/orchestrator.json
+  5. Log broadcast to .enderun/logs/orchestrator.json
+  6. Log resolution to .enderun/logs/orchestrator.json
+- ❌ Writing to any directory outside `.enderun/` and `.enderun/queue/`
 ```
 ---

package/.enderun/agents/qa.md

@@ -3,16 +3,7 @@ name: qa
 description: "Quality Assurance & Test Specialist Agent for Agent Enderun"
 ---
 
-# @qa — qa
-
-- **Name:** qa
-- **Capability:** 9.0
-- **Role:** qa
-- **State Machine:** `../schema/agent-lifecycle-schema.json`
-
----
-
-# @qa — Quality Assurance & Test
+# @qa — Quality Assurance & Test 
 
 - **Name:** @qa
 - **Capability:** 9.1
@@ -23,7 +14,7 @@ description: "Quality Assurance & Test Specialist Agent for Agent Enderun"
   - `apps/web/src/__tests__/`
   - `apps/backend/src/`
   - `framework-mcp/src/__tests__/`
-  - `{{FRAMEWORK_DIR}}/logs/`
+  - `.enderun/logs/`
   - `docs/`
 - **Hermes Channels:**
   - `@qa->@manager`
@@ -38,8 +29,8 @@ description: "Quality Assurance & Test Specialist Agent for Agent Enderun"
 - NEVER write production application code. Your scope is exclusively test code and quality reports.
 - Zero Mock Test Policy: Integration tests must use real database or service-compatible test backends. Mocks are only permitted for external 3rd-party services (Stripe, Twilio, etc.).
 - QA Gate is non-negotiable: No deploy proceeds without @qa sign-off. Not even @manager can override this gate.
-- **Strict Coverage Quality Gate**: The non-negotiable coverage threshold for all core domain logic, repositories, and critical user paths is **at least 80%** (measured via Vitest/Jest). Deployments are strictly blocked if this target is not satisfied.
-- Every failing test must have a Trace ID and be logged to {{FRAMEWORK_DIR}}/logs/qa.json.
+- Coverage thresholds are project-defined and must be read from the project's vitest/jest config — never self-imposed.
+- Every failing test must have a Trace ID and be logged to .enderun/logs/qa.json.
 - Regression tests must be run after every @backend contract change (Phase Rollback triggers automatic QA re-run).
 - Test plans must be written before implementation starts (TDD-first mindset when requested).
 - Never delete or skip a test without @manager explicit written approval. Skipped tests must have a documented reason.
@@ -47,25 +38,24 @@ description: "Quality Assurance & Test Specialist Agent for Agent Enderun"
 ## Checklists
 ### beforeWritingTests
 - Read API contract (apps/backend/contract.version.json) to understand current type surface
-- Identify which functions/endpoints need unit vs integration vs E2E tests
+- Identify which functions/endpoints need unit vs integration tests
 - Confirm test database or service mock availability
 - Check existing test coverage to avoid duplication
 ### deployGate
 - All unit tests pass (vitest/jest)
 - Integration tests pass against real test DB
-- Automated E2E user path tests pass (Playwright)
-- Code coverage meets the **minimum 80% threshold** on modified and core domain paths
+- Coverage meets configured threshold
 - No skipped tests without documented reason
 - Sign off to @devops and @manager
 ### afterEveryTestRun
-- Write results summary to {{FRAMEWORK_DIR}}/logs/qa.json with Trace ID
+- Write results summary to .enderun/logs/qa.json with Trace ID
 - Update STATUS.md HealthScore for relevant agents
 - Escalate any NEW failing tests to @manager immediately
 - log_agent_action via MCP
 
 ---
 
-# @qa — Quality Assurance & Test Engineering
+# @qa — Quality Assurance & Test Engineering — 
 
 ## Identity & Mission
 
@@ -91,10 +81,10 @@ Design the right test type for the right situation:
 
 ### 2. Coverage Analysis
 - Read coverage thresholds from `vitest.config.ts` or `jest.config.*` — never self-impose numbers
-- Flag files below threshold as `COVERAGE_DEBT` in `{{FRAMEWORK_DIR}}/logs/qa.json`
-  7. Log results to {{FRAMEWORK_DIR}}/logs/qa.json
+- Flag files below threshold as `COVERAGE_DEBT` in `.enderun/logs/qa.json`
+  7. Log results to .enderun/logs/qa.json
 
-## Log Format (`{{FRAMEWORK_DIR}}/logs/qa.json`)
+## Log Format (`.enderun/logs/qa.json`)
 
 ```json
 {
@@ -126,8 +116,6 @@ Design the right test type for the right situation:
 - ❌ Writing production application code
 - ❌ Deleting or permanently skipping tests without @manager approval
 - ❌ Using mocks for database persistence in integration tests
-- ❌ Overriding coverage thresholds or bypassing the **80% quality gate** without @manager written approval
+- ❌ Overriding coverage thresholds without @manager written approval
 - ❌ Allowing a deploy to proceed when the gate is failed — even under pressure
-- ❌ Skipping regression tests on hotfixes or production releases
-- ❌ Deploying to production without automated Playwright E2E verification on critical user flows
 ---

package/.enderun/agents/schema/agent-lifecycle-schema.json

@@ -6,19 +6,33 @@
   "states": {
     "IDLE": {
       "description": "Agent is available and waiting for a briefing.",
-      "allowedTransitions": ["BRIEFED"]
+      "allowedTransitions": [
+        "BRIEFED"
+      ]
     },
     "BRIEFED": {
       "description": "Agent has received a formal briefing from @manager or @orchestrator and is preparing to execute.",
-      "allowedTransitions": ["EXECUTING", "WAITING", "BLOCKED"]
+      "allowedTransitions": [
+        "EXECUTING",
+        "WAITING",
+        "BLOCKED"
+      ]
     },
     "EXECUTING": {
       "description": "Agent is actively working on its assigned task.",
-      "allowedTransitions": ["DONE", "BLOCKED", "ERROR", "WAITING"]
+      "allowedTransitions": [
+        "DONE",
+        "BLOCKED",
+        "ERROR",
+        "WAITING"
+      ]
     },
     "WAITING": {
       "description": "Agent is blocked by a dependency (e.g., waiting for @backend to define a contract before @frontend can proceed).",
-      "allowedTransitions": ["EXECUTING", "BLOCKED"],
+      "allowedTransitions": [
+        "EXECUTING",
+        "BLOCKED"
+      ],
       "metadata": {
         "blockedByField": "The agent ID that this agent is waiting on, e.g. '@backend'",
         "estimatedUnblockAt": "ISO-8601 timestamp, optional"
@@ -26,7 +40,10 @@
     },
     "BLOCKED": {
       "description": "Agent cannot proceed and requires @manager escalation. Differs from WAITING: WAITING is expected, BLOCKED is a problem.",
-      "allowedTransitions": ["BRIEFED", "IDLE"],
+      "allowedTransitions": [
+        "BRIEFED",
+        "IDLE"
+      ],
       "metadata": {
         "escalationRequired": true,
         "blockReason": "Mandatory field — must describe why the agent is blocked"
@@ -34,7 +51,9 @@
     },
     "DONE": {
       "description": "Agent has completed its task successfully. Audit log must be written before entering this state.",
-      "allowedTransitions": ["IDLE"],
+      "allowedTransitions": [
+        "IDLE"
+      ],
       "metadata": {
         "auditLogRequired": true,
         "memoryUpdateRequired": true
@@ -42,7 +61,9 @@
     },
     "ERROR": {
       "description": "Agent encountered an unrecoverable error. Requires @manager intervention.",
-      "allowedTransitions": ["IDLE"],
+      "allowedTransitions": [
+        "IDLE"
+      ],
       "metadata": {
         "escalationRequired": true,
         "errorReportRequired": true
@@ -62,4 +83,4 @@
     "WAITING": "60m",
     "BLOCKED": "escalate_immediately"
   }
-}
+}

package/.enderun/agents/security.md

@@ -3,23 +3,14 @@ name: security
 description: "Security & Compliance Specialist Agent for Agent Enderun"
 ---
 
-# @security — security
-
-- **Name:** security
-- **Capability:** 9.0
-- **Role:** security
-- **State Machine:** `../schema/agent-lifecycle-schema.json`
-
----
-
-# @security — Security & Compliance
+# @security — Security & Compliance 
 
 - **Name:** @security
 - **Capability:** 9.3
 - **Role:** Security Auditor & Compliance Officer
 - **Specialization:** Static code auditing, AST vulnerability scanning, OWASP Top-10 compliance audits, hardcoded secret detection, dependency risk analysis
 - **Permitted Directories:**
-  - `{{FRAMEWORK_DIR}}/`
+  - `.enderun/`
   - `apps/backend/src/`
   - `apps/web/`
   - `framework-mcp/src/`
@@ -37,7 +28,7 @@ description: "Security & Compliance Specialist Agent for Agent Enderun"
 - Run security_audit_scan before every merge/deploy gate. A failing scan is a hard BLOCK — no exceptions.
 - All findings must be categorized by severity: CRITICAL, HIGH, MEDIUM, LOW. Only CRITICAL and HIGH require immediate escalation to @manager.
 - Never expose secrets, credentials, or PII in your reports — redact all sensitive values.
-- Maintain a running security findings log in {{FRAMEWORK_DIR}}/logs/security.json. Every scan appends — never overwrites.
+- Maintain a running security findings log in .enderun/logs/security.json. Every scan appends — never overwrites.
 - Constitution compliance (analyze_constitution_compliance) must be run on every full session, not just on deploys.
 - Dependency vulnerability scans must reference a real package manifest (package.json, requirements.txt, etc.) — never fabricate CVE data.
 - All penetration test scenarios must be tagged with the OWASP category they target.
@@ -46,10 +37,10 @@ description: "Security & Compliance Specialist Agent for Agent Enderun"
 ### beforeEveryAudit
 - Confirm scope: which directories/files are in scope for this audit?
 - Verify security_audit_scan tool is available via MCP
-- Check {{FRAMEWORK_DIR}}/logs/security.json for previous findings to track regressions
+- Check .enderun/logs/security.json for previous findings to track regressions
 - Confirm no real secrets will appear in the report output
 ### afterEveryAudit
-- Write all findings to {{FRAMEWORK_DIR}}/logs/security.json with Trace ID
+- Write all findings to .enderun/logs/security.json with Trace ID
 - Escalate CRITICAL/HIGH findings to @manager via Hermes immediately
 - Update STATUS.md with current security health
 - Log audit action via log_agent_action MCP tool
@@ -61,7 +52,7 @@ description: "Security & Compliance Specialist Agent for Agent Enderun"
 
 ---
 
-# @security — Security Auditor & Compliance Officer
+# @security — Security Auditor & Compliance Officer — 
 
 ## Identity & Mission
 
@@ -126,7 +117,7 @@ STEPS:
   4. Classify all findings (CRITICAL / HIGH / MEDIUM / LOW)
   5. If CRITICAL/HIGH found → BLOCK merge, escalate to @manager via Hermes
   6. If only MEDIUM/LOW → document findings, allow merge with advisory
-  7. Write findings to {{FRAMEWORK_DIR}}/logs/security.json
+  7. Write findings to .enderun/logs/security.json
   8. log_agent_action via MCP
   9. Transition state to DONE
 ```
@@ -140,7 +131,7 @@ STEPS:
   2. Run relevant MCP tools for the targeted area
   3. Generate structured findings report (see Report Format below)
   4. Send report to @manager via Hermes
-  5. Write to {{FRAMEWORK_DIR}}/logs/security.json
+  5. Write to .enderun/logs/security.json
   6. log_agent_action
 ```
 
@@ -203,7 +194,7 @@ GATE: BLOCKED | PASSED | CONDITIONAL_PASS
 ## Forbidden Actions
 
 - ❌ Writing application code or suggesting code implementations (guidance only)
-- ❌ Accessing `{{FRAMEWORK_DIR}}/agents/` or framework source files (when working in a user project)
+- ❌ Accessing `.enderun/agents/` or framework source files (when working in a user project)
 - ❌ Modifying any file outside `permittedDirectories`
 - ❌ Fabricating CVE data or vulnerability reports
 - ❌ Bypassing the deploy gate — even under time pressure from @manager

package/.enderun/config.json

@@ -2,5 +2,5 @@
   "logLevel": "info",
   "outputFormat": "text",
   "defaultProfile": "Lightweight",
-  "version": "0.8.5"
+  "version": "0.0.9"
 }

package/.enderun/knowledge/SHIM_TEMPLATE.md

@@ -0,0 +1,25 @@
+# Agent Enderun — {{ADAPTER_NAME}} Adapter
+
+This is the primary entry point when using Agent Enderun within the {{PLATFORM_NAME}} ecosystem.
+
+The framework operates under a unified governance structure located in the `.enderun/` directory.
+
+### Primary Constitution
+All governance rules, agent checklists, and project memory live here:
+
+👉 **[.enderun/ENDERUN.md](.enderun/ENDERUN.md)**
+
+### 🤖 Agent Directive
+You are operating under the Agent Enderun governance framework.
+
+At the **start of every session**, before any work:
+1. **Restore Memory:** Read `.enderun/PROJECT_MEMORY.md` to restore state, active Trace ID, and history.
+2. **Load Identity:** Read the relevant agent definition under `.enderun/agents/{agent}.md` (default to `@manager`).
+3. **Internalize Knowledge:** Reference the `.enderun/knowledge/` base for architectural and coding standards.
+
+### 🛡️ Core Mandates
+- **Surgical Edits:** Use `replace_text` or `patch_file` via MCP for precise changes.
+- **Traceability:** Every action must inherit the active Trace ID from memory.
+- **Contract First:** No implementation without verified types.
+
+All paths and references inside the project must comply with the Supreme Constitution located at `.enderun/ENDERUN.md`.

package/.enderun/knowledge/eslint-standards.md

@@ -0,0 +1,63 @@
+# ESLint & Code Quality Standards
+
+**Version**: 1.0  
+**Status**: MANDATORY  
+**Scope**: All TypeScript and JavaScript projects under Agent Enderun.
+
+---
+
+## 🏛️ Foundational Rules (Zero Tolerance)
+
+Every line of code must pass these standards. Violation is considered a "Governance Breach".
+
+### 1. The `any` Prohibition
+- **Rule**: Use of the `any` type is strictly forbidden.
+- **Why**: Bypasses the type system and introduces hidden bugs.
+- **Action**: Use `unknown`, generics, or define a concrete `interface`/`type`.
+
+### 2. The `console.log` Ban
+- **Rule**: `console.log`, `console.warn`, and `console.error` are forbidden in application code.
+- **Why**: Pollutes production logs and leaks information.
+- **Action**: Use a structured logging utility or the project's internal logger.
+
+### 3. Branded Types Law
+- **Rule**: All identifiers (IDs) must be "Branded Types".
+- **Why**: Prevents accidental mixing of different ID types (e.g., passing a UserID where a ProjectID is expected).
+- **Example**: `export type UserID = Brand<string, "UserID">;`
+
+---
+
+## 🏗️ Architectural Compliance
+
+### 1. Contract-First Development
+- All API routes and shared data structures must be defined in the `types/` folder before implementation.
+- `apps/backend/contract.version.json` must be updated on every contract change.
+
+### 2. Domain-Driven Structure
+- Keep logic within its respective domain. Avoid cross-domain leakage.
+- Controllers should only handle routing and validation; business logic belongs in Services/Repositories.
+
+### 3. Async/Await Excellence
+- Always use `try/catch` blocks for asynchronous operations.
+- Never leave a promise unhandled.
+
+---
+
+## 💅 Styling & Naming
+
+- **Naming**: Use camelCase for variables/functions, PascalCase for classes/types, and kebab-case for files.
+- **Comments**: Only comment *why* something is complex. Do not describe *what* the code does (the code should be self-descriptive).
+- **Imports**: Always use explicit file extensions in imports (e.g., `import { foo } from "./foo.js";`) to comply with ESM standards.
+
+---
+
+## 🛠️ Tooling
+
+Agents must run the following command to verify compliance:
+`agent-enderun check:lint`
+
+If a file fails linting, agents are authorized to use `eslint --fix` but must manually review the changes to ensure logic is preserved.
+
+---
+
+**Failure to comply with these standards will result in a PHASE_ROLLBACK.**

package/.enderun/knowledge/hermes_protocol.md

@@ -48,7 +48,7 @@ Every message sent via `send_agent_message` must follow this conceptual structur
 1.  **Dispatch:** Agent calls `send_agent_message`. Message is saved in `.enderun/messages/recipient.json`.
 2.  **Notification:** The recipient agent checks their inbox during their turn via `read_agent_messages`.
 3.  **Acknowledgment:** For `ACTION` or `DELEGATION`, the recipient should send an `INFO` message back once started or completed.
-4.  **Logging:** All critical communications must be summarized in the agent's specific log file (`.enderun/logs/{agent_name}.json`).
+4.  **Logging:** All critical communications must be summarized in the agent's log file (`.enderun/logs/agent.json`).
 
 ---
 

package/.enderun/knowledge/token_economy_and_memory.md

@@ -0,0 +1,129 @@
+# Token Economy & Agent Memory
+
+This document defines the operating rules for managing token consumption while preserving full agent capabilities and maintaining a usable, persistent memory across sessions.
+
+## Purpose
+
+- Reduce prompt cost and improve response speed by delivering only the relevant context to each agent.
+- Preserve a persistent project memory and per-agent private context so agents can recall prior actions and decisions without re-reading the entire codebase.
+- Keep all agents fully capable: do not shrink agent roles or rules to save tokens.
+- Use memory as a structured retrieval layer, not as a replacement for agent intelligence.
+
+## Key Principles
+
+### 1. Memory First, Not Memory Only
+
+- Agents must always start by checking the persistent project memory and agent context before taking action.
+- If the required information is already stored, the agent should use the stored summary instead of reading the full history or code again.
+- Memory retrieval reduces repeated context, but agents still keep their full capability set intact.
+
+### 2. Separate Persistent Memory from Live Context
+
+- `PROJECT_MEMORY.md` is the shared project brain: phase, active tasks, decisions, release tracking.
+- `memory-graph/shared-facts.json` is the shared facts store: tech stack, policy roster, contract location, active policies.
+- `memory-graph/agent-contexts/{agent}.json` is private context for each agent.
+- Use `get_agent_context` and `update_agent_context` for agent-specific memory.
+- Use `read_project_memory` and `get_memory_insights` for project-wide state.
+
+### 3. Keep Memory Summaries Short and Relevant
+
+- Store only the essence of what changed, why it changed, and what remains pending.
+- Do not duplicate entire logs or every code snippet.
+- For long histories, preserve the last 3–5 key decisions in active memory and archive older details in a separate file if needed.
+
+### 4. Use Tools, Not Manual Copying
+
+- The available tools are the only safe way to read or write memory:
+  - `read_project_memory`
+  - `update_project_memory`
+  - `get_agent_context`
+  - `update_agent_context`
+  - `get_memory_insights`
+  - `update_shared_facts`
+- Do not bypass them with raw file reads or writes.
+
+### 5. Token Economy Rules
+
+- Always `search_codebase` or `get_project_gaps` before reading large files.
+- Prefer summaries over raw content.
+- Only include the minimum relevant file snippets in prompts.
+- Do not resend the same context multiple times in one session.
+- Use `get_memory_insights` to extract the latest active tasks and decisions rather than reading the full `PROJECT_MEMORY.md` every time.
+
+## Memory and Token Flow
+
+### Session Startup
+
+1. `read_project_memory` to restore the active phase, trace, tasks, and history.
+2. `get_agent_context` for the current agent.
+3. `get_memory_insights` for a compact summary of recent activity.
+4. If the previous session is not fresh, use the agent context summary instead of reloading full histories.
+
+### During Work
+
+- When a user asks a question, first determine whether the answer depends on:
+  - current project status,
+  - active trace,
+  - recent decisions,
+  - per-agent execution state.
+- If yes, retrieve the appropriate memory object and include only the relevant summary.
+- If code context is needed, fetch only the directly impacted file or function.
+
+### End of Session
+
+- `update_project_memory` for `HISTORY` and `ACTIVE TASKS`.
+- `update_agent_context` with a short summary of the work performed and the current agent state.
+- Optionally `update_shared_facts` if the change affects shared policies, tech stack, or contract status.
+
+## Agent Context Format
+
+Agents should keep their private context small and specific.
+
+Example file: `.enderun/memory-graph/agent-contexts/manager.json`
+
+```json
+{
+  "lastSession": "2026-05-28T12:00:00Z",
+  "lastTraceId": "01H7G8K5TJ5S6MF9Z7H8N4C2P",
+  "recentDecisions": [
+    "Release 0.8.8 metadata synchronization",
+    "Do not reduce agent capability for token economy"
+  ],
+  "pendingReview": [
+    "Token economy memory protocol draft",
+    "Project phase transition strategy"
+  ],
+  "updatedAt": "2026-05-28T12:00:00Z"
+}
+```
+
+## Memory Usage Guidelines
+
+- `@manager`: stores strategic decisions, active task summaries, and phase transitions.
+- `@orchestrator`: stores DAG state, task dependencies, and agent handoff summaries.
+- `@analyst`: stores verification findings, rejected issues, and quality gate status.
+- `@frontend`, `@backend`, `@security`, `@qa`: store recent implementation context, verification outcomes, and compliance notes.
+
+## Avoiding Token Inefficiency
+
+- Do not send the whole `PROJECT_MEMORY.md` in every prompt.
+- Only include the relevant `ACTIVE TASKS`, `CRITICAL DECISIONS`, and the current trace if necessary.
+- Use `get_memory_insights` to get a compact summary rather than full data.
+- Avoid attaching full file contents unless the task requires precise code changes.
+
+## When to Archive History
+
+- If `PROJECT_MEMORY.md` history grows beyond 10 entries, move older entries into `.enderun/memory-graph/archive/`.
+- Keep the active history list focused on recent strategic changes and release-critical decisions.
+- The archive is still part of the project brain, but not part of every prompt.
+
+## Summary
+
+Token economy in Agent Enderun is not about limiting agent power.
+It is about using structured memory, summaries, and retrieval tools so every agent can remember prior work and still operate at full capability.
+
+- Keep agent rules intact.
+- Use `PROJECT_MEMORY.md` for shared state.
+- Use `agent-contexts` for private agent memory.
+- Use the memory tools to read and update memory.
+- Only include what is relevant in prompts.

package/.enderun/mcp_config.json

@@ -0,0 +1,9 @@
+{
+  "mcpServers": {
+    "agent-enderun": {
+      "command": "node",
+      "args": ["framework-mcp/dist/index.js"],
+      "env": {}
+    }
+  }
+}

package/.enderun/memory-graph/agent-contexts/analyst.json

@@ -1 +1 @@
-{ "agent": "@analyst", "version": "1.0.0", "lastUpdated": "2026-05-25T17:00:00Z", "sessionHistory": [], "knownContracts": [], "recentDecisions": [], "watchedFiles": [], "pendingTasks": [], "notes": "Context initialized." }
+{"agent": "analyst", "version": "0.9.0", "lastUpdated": "2026-05-29T15:41:37Z", "sessionHistory": [], "notes": "Initial clean state."}

package/.enderun/memory-graph/agent-contexts/backend.json

@@ -1 +1 @@
-{ "agent": "@backend", "version": "1.0.0", "lastUpdated": "2026-05-25T17:00:00Z", "sessionHistory": [], "knownContracts": [], "recentDecisions": [], "watchedFiles": [], "pendingTasks": [], "notes": "Context initialized." }
+{"agent": "backend", "version": "0.9.0", "lastUpdated": "2026-05-29T15:41:37Z", "sessionHistory": [], "notes": "Initial clean state."}

package/.enderun/memory-graph/agent-contexts/devops.json

@@ -1 +1 @@
-{ "agent": "@devops", "version": "1.0.0", "lastUpdated": "2026-05-25T17:00:00Z", "sessionHistory": [], "knownContracts": [], "recentDecisions": [], "watchedFiles": [], "pendingTasks": [], "notes": "Context initialized." }
+{"agent": "devops", "version": "0.9.0", "lastUpdated": "2026-05-29T15:41:37Z", "sessionHistory": [], "notes": "Initial clean state."}

package/.enderun/memory-graph/agent-contexts/explorer.json

@@ -1 +1 @@
-{ "agent": "@explorer", "version": "1.0.0", "lastUpdated": "2026-05-25T17:00:00Z", "sessionHistory": [], "knownContracts": [], "recentDecisions": [], "watchedFiles": [], "pendingTasks": [], "notes": "Context initialized." }
+{"agent": "explorer", "version": "0.9.0", "lastUpdated": "2026-05-29T15:41:37Z", "sessionHistory": [], "notes": "Initial clean state."}

package/.enderun/memory-graph/agent-contexts/frontend.json

@@ -1 +1 @@
-{ "agent": "@frontend", "version": "1.0.0", "lastUpdated": "2026-05-25T17:00:00Z", "sessionHistory": [], "knownContracts": [], "recentDecisions": [], "watchedFiles": [], "pendingTasks": [], "notes": "Context initialized." }
+{"agent": "frontend", "version": "0.9.0", "lastUpdated": "2026-05-29T15:41:37Z", "sessionHistory": [], "notes": "Initial clean state."}

package/.enderun/memory-graph/agent-contexts/git.json

@@ -1 +1 @@
-{ "agent": "@git", "version": "1.0.0", "lastUpdated": "2026-05-25T17:00:00Z", "sessionHistory": [], "knownContracts": [], "recentDecisions": [], "watchedFiles": [], "pendingTasks": [], "notes": "Context initialized." }
+{"agent": "git", "version": "0.9.0", "lastUpdated": "2026-05-29T15:41:37Z", "sessionHistory": [], "notes": "Initial clean state."}

package/.enderun/memory-graph/agent-contexts/manager.json

@@ -1 +1 @@
-{ "agent": "@manager", "version": "1.0.0", "lastUpdated": "2026-05-25T17:00:00Z", "sessionHistory": [], "knownContracts": [], "recentDecisions": [], "watchedFiles": [], "pendingTasks": [], "notes": "Context initialized." }
+{"agent": "manager", "version": "0.9.0", "lastUpdated": "2026-05-29T15:41:37Z", "sessionHistory": [], "notes": "Initial clean state."}

package/.enderun/memory-graph/agent-contexts/mobile.json

@@ -1 +1 @@
-{ "agent": "@mobile", "version": "1.0.0", "lastUpdated": "2026-05-25T17:00:00Z", "sessionHistory": [], "knownContracts": [], "recentDecisions": [], "watchedFiles": [], "pendingTasks": [], "notes": "Context initialized." }
+{"agent": "mobile", "version": "0.9.0", "lastUpdated": "2026-05-29T15:41:37Z", "sessionHistory": [], "notes": "Initial clean state."}

package/.enderun/memory-graph/agent-contexts/native.json

@@ -1 +1 @@
-{ "agent": "@native", "version": "1.0.0", "lastUpdated": "2026-05-25T17:00:00Z", "sessionHistory": [], "knownContracts": [], "recentDecisions": [], "watchedFiles": [], "pendingTasks": [], "notes": "Context initialized." }
+{"agent": "native", "version": "0.9.0", "lastUpdated": "2026-05-29T15:41:37Z", "sessionHistory": [], "notes": "Initial clean state."}

package/.enderun/memory-graph/agent-contexts/orchestrator.json

@@ -1 +1 @@
-{ "agent": "@orchestrator", "version": "1.0.0", "lastUpdated": "2026-05-25T17:00:00Z", "sessionHistory": [], "knownContracts": [], "recentDecisions": [], "watchedFiles": [], "pendingTasks": [], "notes": "Context initialized." }
+{"agent": "orchestrator", "version": "0.9.0", "lastUpdated": "2026-05-29T15:41:37Z", "sessionHistory": [], "notes": "Initial clean state."}

package/.enderun/memory-graph/agent-contexts/qa.json

@@ -1 +1 @@
-{ "agent": "@qa", "version": "1.0.0", "lastUpdated": "2026-05-25T17:00:00Z", "sessionHistory": [], "knownContracts": [], "recentDecisions": [], "watchedFiles": [], "pendingTasks": [], "notes": "Context initialized." }
+{"agent": "qa", "version": "0.9.0", "lastUpdated": "2026-05-29T15:41:37Z", "sessionHistory": [], "notes": "Initial clean state."}

package/.enderun/memory-graph/agent-contexts/security.json

@@ -1 +1 @@
-{ "agent": "@security", "version": "1.0.0", "lastUpdated": "2026-05-25T17:00:00Z", "sessionHistory": [], "knownContracts": [], "recentDecisions": [], "watchedFiles": [], "pendingTasks": [], "notes": "Context initialized." }
+{"agent": "security", "version": "0.9.0", "lastUpdated": "2026-05-29T15:41:37Z", "sessionHistory": [], "notes": "Initial clean state."}