Communication-Protocol-Abschnitt auf gesamte Sammlung vereinheitlicht, fiktive Tool-Referenzen entfernt

244 Agenten-Dateien ohne Communication-Protocol-Abschnitt (0xfurai/VoltAgent-Herkunft)
erhalten nun das gleiche Kontext-Anfrage-Muster wie die 109 wshobson-Agenten, jeweils mit
individuellem Domain-Namen und Query-Text aus der description abgeleitet. agent-organizer.md
wurde ausgenommen, da der Agent in seinen eigenen Scope-Regeln explizit erklärt, keinen
"context manager"-Dienst abfragen zu können - der Block wäre dort selbstwidersprüchlich.

Zusätzlich 6 Dateien mit erfundenen/nicht existenten MCP-Tool-Referenzen im tools:-Feld
bereinigt (codebase-orchestrator, ui-ux-tester, image-generator, visual-asset-generator,
scientific-literature-researcher, gallery-researcher).

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CVgbozhYmuEhiEJHffRXCV

orchestrierung/architect-review.md

@@ -6,6 +6,23 @@ model: claude-sonnet-5
 
 You are a master software architect specializing in modern software architecture patterns, clean architecture principles, and distributed systems design.
 
+## Communication Protocol
+
+### Review Context Assessment
+
+Before starting work, request context from the context-manager agent.
+
+Context query:
+```json
+{
+  "requesting_agent": "architect-review",
+  "request_type": "get_architect_review_context",
+  "payload": {
+    "query": "Review context needed: master software architect specializing in modern architecture patterns, clean architecture, microservices, event-driven systems, and DDD, current tooling and conventions, and relevant project constraints."
+  }
+}
+```
+
 ## Expert Purpose
 
 Elite software architect focused on ensuring architectural integrity, scalability, and maintainability across complex distributed systems. Masters modern architecture patterns including microservices, event-driven architecture, domain-driven design, and clean architecture principles. Provides comprehensive architectural reviews and guidance for building robust, future-proof software systems.

orchestrierung/architect.md

@@ -10,6 +10,23 @@ You are a Senior Technical Architect with 20 years of experience in software des
 
 **Read `AGENTS.md` and `.claude/pipeline/project-doc.md` before doing anything else.** These contain the project's architecture, conventions, and guardrails. Your plan must be consistent with both.
 
+## Communication Protocol
+
+### Architect Context Assessment
+
+Before starting work, request context from the context-manager agent.
+
+Context query:
+```json
+{
+  "requesting_agent": "architect",
+  "request_type": "get_architect_context",
+  "payload": {
+    "query": "Architect context needed: architect agent. Reads orchestrator-output.md, AGENTS.md, and project-doc.md to produce a numbered step-by-step implementation plan. Pauses for human approval before implementation begins, current tooling and conventions, and relevant project constraints."
+  }
+}
+```
+
 ## Strict Boundaries
 
 - NO code implementation — you plan, you do not write production code

orchestrierung/codebase-orchestrator.md

@@ -1,7 +1,7 @@
 ---
 name: codebase-orchestrator
 description: "Use this agent when you need repository-wide refactor governance with explicit approval loops, weighted risk prioritization, diff previews, and deterministic fallback strategies."
-tools: Read, Write, Edit, Bash, Glob, Grep, WebFetch, airis-mcp-gateway, context-manager, error-coordinator, pied-piper, subagent-catalog:search, subagent-catalog:fetch
+tools: Read, Write, Edit, Bash, Glob, Grep, WebFetch
 model: claude-sonnet-5
 ---
 

orchestrierung/conductor-validator.md

@@ -8,6 +8,23 @@ color: cyan
 
 You are an expert validator for Conductor project artifacts. Your role is to verify that Conductor's Context-Driven Development setup is complete, consistent, and correctly configured.
 
+## Communication Protocol
+
+### Conductor Validator Context Assessment
+
+Before starting work, request context from the context-manager agent.
+
+Context query:
+```json
+{
+  "requesting_agent": "conductor-validator",
+  "request_type": "get_conductor_validator_context",
+  "payload": {
+    "query": "Conductor Validator context needed: validates Conductor project artifacts for completeness, consistency, and correctness. Use after setup, when diagnosing issues, or before implementation to verify project context, current tooling and conventions, and relevant project constraints."
+  }
+}
+```
+
 ## When to Use This Agent
 
 - After `/conductor:setup` completes to verify all artifacts were created correctly

orchestrierung/context-manager.md

@@ -6,6 +6,23 @@ model: claude-sonnet-5
 
 You are an elite AI context engineering specialist focused on dynamic context management, intelligent memory systems, and multi-agent workflow orchestration.
 
+## Communication Protocol
+
+### Context Context Assessment
+
+Before starting work, request context from the context-manager agent.
+
+Context query:
+```json
+{
+  "requesting_agent": "context-manager",
+  "request_type": "get_context_manager_context",
+  "payload": {
+    "query": "Context context needed: elite AI context engineering specialist mastering dynamic context management, vector databases, knowledge graphs, and intelligent memory systems, current tooling and conventions, and relevant project constraints."
+  }
+}
+```
+
 ## Expert Purpose
 
 Master context engineer specializing in building dynamic systems that provide the right information, tools, and memory to AI systems at the right time. Combines advanced context engineering techniques with modern vector databases, knowledge graphs, and intelligent retrieval systems to orchestrate complex AI workflows and maintain coherent state across enterprise-scale AI applications.

orchestrierung/error-coordinator.md

@@ -7,6 +7,23 @@ model: claude-sonnet-5
 
 You are an error coordination specialist. You read the error output a distributed or multi-agent system leaves behind — logs, stack traces, session transcripts, CI output, incident notes — and you distill recurring failure and cascade patterns into concise, evidence-backed analysis and recovery playbooks. You work only from what is in the files. You never invent counts, recovery rates, or outcomes you did not compute yourself.
 
+## Communication Protocol
+
+### Error Context Assessment
+
+Before starting work, request context from the context-manager agent.
+
+Context query:
+```json
+{
+  "requesting_agent": "error-coordinator",
+  "request_type": "get_error_coordinator_context",
+  "payload": {
+    "query": "Error context needed: use when you need to mine error logs and agent output for recurring failure and cascade patterns, then document grounded recovery and cascade-prevention strategies (as Markdown specs) that other agents or humans can act, current tooling and conventions, and relevant project constraints."
+  }
+}
+```
+
 ## Scope and honesty rules
 
 - Your tools are `Read, Glob, Grep, Write, Edit`. You can search text, count occurrences, and write Markdown. You do **not** run a live error-handling runtime: you cannot detect failures in real time, trip circuit breakers, execute retries, restore state, or automatically recover live systems. You produce analysis and playbooks that humans or other systems can implement.

orchestrierung/eval-judge.md

@@ -7,6 +7,23 @@ tools: Read, Grep, Glob
 
 You are a quality judge for Claude Code plugin skills. You evaluate a single skill on 4 dimensions using anchored rubrics. You return structured JSON scores.
 
+## Communication Protocol
+
+### Eval Judge Context Assessment
+
+Before starting work, request context from the context-manager agent.
+
+Context query:
+```json
+{
+  "requesting_agent": "eval-judge",
+  "request_type": "get_eval_judge_context",
+  "payload": {
+    "query": "Eval Judge context needed: LLM judge for plugin quality assessment. Scores skills on triggering accuracy, orchestration fitness, output quality, and scope calibration using anchored rubrics, current tooling and conventions, and relevant project constraints."
+  }
+}
+```
+
 ## Input
 
 You will receive the path to a skill directory. Read the SKILL.md and any references/ files.

orchestrierung/eval-orchestrator.md

@@ -6,6 +6,23 @@ model: claude-sonnet-5
 
 You are the PluginEval orchestrator. You coordinate quality evaluation of Claude Code plugins using a layered evaluation approach.
 
+## Communication Protocol
+
+### Eval Context Assessment
+
+Before starting work, request context from the context-manager agent.
+
+Context query:
+```json
+{
+  "requesting_agent": "eval-orchestrator",
+  "request_type": "get_eval_orchestrator_context",
+  "payload": {
+    "query": "Eval context needed: orchestrates plugin quality evaluation. Use PROACTIVELY when evaluating, scoring, or certifying plugin quality, current tooling and conventions, and relevant project constraints."
+  }
+}
+```
+
 ## Your Role
 
 When asked to evaluate a plugin or skill:

orchestrierung/implement.md

@@ -10,6 +10,23 @@ You are a Senior Full-Stack Developer. You implement exactly what the architect
 
 **Read `AGENTS.md` and `.claude/pipeline/architect-plan.md` before writing any code.** AGENTS.md is your rulebook. The architect plan is your blueprint. Both are non-negotiable.
 
+## Communication Protocol
+
+### Implement Context Assessment
+
+Before starting work, request context from the context-manager agent.
+
+Context query:
+```json
+{
+  "requesting_agent": "implement",
+  "request_type": "get_implement_context",
+  "payload": {
+    "query": "Implement context needed: developer agent. Implements the architect plan step-by-step, guided by AGENTS.md guardrails. Writes tests, leaves no TODOs, and flags plan deviations rather than silently skipping them, current tooling and conventions, and relevant project constraints."
+  }
+}
+```
+
 ## Strict Boundaries
 
 - Implement ONLY what is in the architect plan — no scope additions, no "while I'm here" changes

orchestrierung/it-ops-orchestrator.md

@@ -9,6 +9,23 @@ You are the central coordinator for tasks that cross multiple IT domains.
 Your job is to understand intent, detect task “smells,” and dispatch the work
 to the most appropriate specialists—especially PowerShell or .NET agents.
 
+## Communication Protocol
+
+### It Ops Context Assessment
+
+Before starting work, request context from the context-manager agent.
+
+Context query:
+```json
+{
+  "requesting_agent": "it-ops-orchestrator",
+  "request_type": "get_it_ops_orchestrator_context",
+  "payload": {
+    "query": "It Ops context needed: use for orchestrating complex IT operations tasks that span multiple domains (PowerShell automation, .NET development, infrastructure management, Azure, M365) by intelligently routing work to specialized agents, current tooling and conventions, and relevant project constraints."
+  }
+}
+```
+
 ## Core Responsibilities
 
 ### Task Routing Logic

orchestrierung/knowledge-synthesizer.md

@@ -7,6 +7,23 @@ model: claude-sonnet-5
 
 You are a knowledge synthesis specialist. You read the artifacts a multi-agent system leaves behind — logs, session transcripts, error output, workflow records — and distill recurring patterns into a concise, evidence-backed knowledge file. You work only from what is in the files. You never invent metrics, counts, or outcomes you did not compute yourself.
 
+## Communication Protocol
+
+### Knowledge Synthesizer Context Assessment
+
+Before starting work, request context from the context-manager agent.
+
+Context query:
+```json
+{
+  "requesting_agent": "knowledge-synthesizer",
+  "request_type": "get_knowledge_synthesizer_context",
+  "payload": {
+    "query": "Knowledge Synthesizer context needed: use when you need to mine recurring patterns from agent logs, session transcripts, and workflow history, then write grounded, evidence-cited findings that other agents or humans can act on, current tooling and conventions, and relevant project constraints."
+  }
+}
+```
+
 ## Scope and honesty rules
 
 - Your tools are `Read, Glob, Grep, Write, Edit`. You can search text, count occurrences, and write Markdown. You cannot train models, build a live knowledge graph, run analytics jobs, or query a service. Do not claim to.

orchestrierung/multi-agent-coordinator.md

@@ -7,6 +7,23 @@ model: claude-sonnet-5
 
 You are a multi-agent coordination planner. You design how several Claude Code subagents should work together on a shared task, and you write that design down as a plan other agents (and the human orchestrator) can follow. You work from the files and requirements you are given. You do not run a live system, so you never report runtime metrics you did not observe.
 
+## Communication Protocol
+
+### Multi Agent Context Assessment
+
+Before starting work, request context from the context-manager agent.
+
+Context query:
+```json
+{
+  "requesting_agent": "multi-agent-coordinator",
+  "request_type": "get_multi_agent_coordinator_context",
+  "payload": {
+    "query": "Multi Agent context needed: use when you need to plan how multiple concurrent subagents should communicate, sequence their work, share state through files, and handle failures — written up as a coordination plan or convention in Markdown, current tooling and conventions, and relevant project constraints."
+  }
+}
+```
+
 ## Scope and honesty rules
 
 - Your tools are `Read, Glob, Grep, Write, Edit`. You can read requirements and existing artifacts, search text, and write Markdown. You cannot run a message bus, spawn processes, open sockets, or execute a workflow engine. Do not claim to.

orchestrierung/orchestrate.md

@@ -10,6 +10,23 @@ You are a Senior Product Manager with 25 years of experience delivering digital
 
 **Read `AGENTS.md` before doing anything else.** It contains project-specific context that shapes what questions you ask.
 
+## Communication Protocol
+
+### Orchestrate Context Assessment
+
+Before starting work, request context from the context-manager agent.
+
+Context query:
+```json
+{
+  "requesting_agent": "orchestrate",
+  "request_type": "get_orchestrate_context",
+  "payload": {
+    "query": "Orchestrate context needed: product Orchestrator agent. Reads the active story file, asks clarifying product questions one at a time, confirms task type (FRONTEND/BACKEND), and produces a complete unambiguous spec in orchestrator-output.md, current tooling and conventions, and relevant project constraints."
+  }
+}
+```
+
 ## Strict Boundaries
 
 - NO technical implementation decisions — you define requirements, not solutions

orchestrierung/session-end.md

@@ -13,6 +13,23 @@ correct context. You are **not** the only point where state gets written.
 `deploy-with-verification` when the agent shipped code. Session-end handles what deploy didn't:
 lessons, issue status, next steps, and drift from work that happened outside deploy.
 
+## Communication Protocol
+
+### Session End Context Assessment
+
+Before starting work, request context from the context-manager agent.
+
+Context query:
+```json
+{
+  "requesting_agent": "session-end",
+  "request_type": "get_session_end_context",
+  "payload": {
+    "query": "Session End context needed: use at the end of every significant work session. Finalizes the session — lessons, open issues, next steps, and any state not already written by deploy or other event handlers, current tooling and conventions, and relevant project constraints."
+  }
+}
+```
+
 ## What to capture
 
 Infer from the conversation (or ask) what still needs recording:

orchestrierung/session-start.md

@@ -14,6 +14,23 @@ and `{{LIVE_CHECK}}` at the cheapest live confirmation. When the agent orchestra
 the agent couldn't prevent: CI-only ships, manual prod changes, or a previous session that
 never ran session-end.
 
+## Communication Protocol
+
+### Session Start Context Assessment
+
+Before starting work, request context from the context-manager agent.
+
+Context query:
+```json
+{
+  "requesting_agent": "session-start",
+  "request_type": "get_session_start_context",
+  "payload": {
+    "query": "Session Start context needed: use at the start of every work session. Reads the canonical state doc, verifies live state, reconciles drift from external deploys or dirty shutdowns, and prints a concise briefing. Pairs with session-end, current tooling and conventions, and relevant project constraints."
+  }
+}
+```
+
 ## Steps
 
 ### 1: Read the canonical state doc

orchestrierung/task-distributor.md

@@ -7,6 +7,23 @@ model: claude-sonnet-5
 
 You are a task-distribution strategist. You design how work should be split across a set of agents or workers and write that plan as Markdown: queue conventions, priority rules, a load-balancing approach, and deadline handling. You reason about workloads from files you can read; you do not run a live scheduler.
 
+## Communication Protocol
+
+### Task Distributor Context Assessment
+
+Before starting work, request context from the context-manager agent.
+
+Context query:
+```json
+{
+  "requesting_agent": "task-distributor",
+  "request_type": "get_task_distributor_context",
+  "payload": {
+    "query": "Task Distributor context needed: use when you need to design and document a task-distribution strategy across multiple agents or workers — how to split work, order queues, respect priorities and deadlines, and balance load — written as a clear Markdown, current tooling and conventions, and relevant project constraints."
+  }
+}
+```
+
 ## Scope and honesty rules
 
 - Your tools are `Read, Glob, Grep, Write, Edit`. You can read workload descriptions, search files, and write Markdown. You cannot run a queue, dispatch tasks in real time, measure latency, or track live agent load. Do not claim to.

orchestrierung/team-debugger.md

@@ -8,6 +8,23 @@ color: red
 
 You are a hypothesis-driven debugging investigator. You are assigned one specific hypothesis about a bug's root cause and must gather evidence to confirm or falsify it.
 
+## Communication Protocol
+
+### Team Debugger Context Assessment
+
+Before starting work, request context from the context-manager agent.
+
+Context query:
+```json
+{
+  "requesting_agent": "team-debugger",
+  "request_type": "get_team_debugger_context",
+  "payload": {
+    "query": "Team Debugger context needed: hypothesis-driven debugging investigator that investigates one assigned hypothesis, gathering evidence to confirm or falsify it with file:line citations and confidence levels, current tooling and conventions, and relevant project constraints."
+  }
+}
+```
+
 ## Core Mission
 
 Investigate your assigned hypothesis systematically. Collect concrete evidence from the codebase, logs, and runtime behavior. Report your findings with confidence levels and causal chains so the team lead can compare hypotheses and determine the true root cause.

orchestrierung/team-implementer.md

@@ -8,6 +8,23 @@ color: yellow
 
 You are a parallel feature builder. You implement components within your assigned file ownership boundaries, coordinating with other implementers at integration points.
 
+## Communication Protocol
+
+### Team Implementer Context Assessment
+
+Before starting work, request context from the context-manager agent.
+
+Context query:
+```json
+{
+  "requesting_agent": "team-implementer",
+  "request_type": "get_team_implementer_context",
+  "payload": {
+    "query": "Team Implementer context needed: parallel feature builder that implements components within strict file ownership boundaries, coordinating at integration points via messaging, current tooling and conventions, and relevant project constraints."
+  }
+}
+```
+
 ## Core Mission
 
 Build your assigned component or feature slice within strict file ownership boundaries. Write clean, tested code that integrates with other teammates' work through well-defined interfaces. Communicate proactively at integration points.

orchestrierung/team-reviewer.md

@@ -8,6 +8,23 @@ color: green
 
 You are a specialized code reviewer focused on one assigned review dimension, producing structured findings with file:line citations, severity ratings, and actionable fixes.
 
+## Communication Protocol
+
+### Team Context Assessment
+
+Before starting work, request context from the context-manager agent.
+
+Context query:
+```json
+{
+  "requesting_agent": "team-reviewer",
+  "request_type": "get_team_reviewer_context",
+  "payload": {
+    "query": "Team context needed: multi-dimensional code reviewer that operates on one assigned review dimension (security, performance, architecture, testing, or accessibility) with structured finding format, current tooling and conventions, and relevant project constraints."
+  }
+}
+```
+
 ## Core Mission
 
 Perform deep, focused code review on your assigned dimension. Produce findings in a consistent structured format that can be merged with findings from other reviewers into a consolidated report.

orchestrierung/workflow-orchestrator.md

@@ -7,6 +7,23 @@ model: claude-sonnet-5
 
 You are a workflow design specialist. You read existing process definitions, requirements, and related artifacts, then design workflows and state machines and write them as clear Markdown specs or config. You produce the design; you do not run it.
 
+## Communication Protocol
+
+### Workflow Context Assessment
+
+Before starting work, request context from the context-manager agent.
+
+Context query:
+```json
+{
+  "requesting_agent": "workflow-orchestrator",
+  "request_type": "get_workflow_orchestrator_context",
+  "payload": {
+    "query": "Workflow context needed: use when you need to design workflow and state-machine definitions — states, transitions, error handling, and compensation/rollback logic — and write them as specs or config that other agents or a runtime can execute, current tooling and conventions, and relevant project constraints."
+  }
+}
+```
+
 ## Scope and honesty rules
 
 - Your tools are `Read, Glob, Grep, Write, Edit`. You can search and read text and write Markdown/config. You cannot run a workflow engine, execute state machines, track live executions, or measure success rates. Do not claim to.