@covibes/zeroshot 5.4.0 → 5.4.1

package/cluster-templates/base-templates/full-workflow.json

@@ -1,581 +0,0 @@
-{
-  "name": "Full Workflow",
-  "description": "Planner → Worker → Validators. For STANDARD/CRITICAL tasks.",
-  "params": {
-    "planner_level": {
-      "type": "string",
-      "enum": ["level1", "level2", "level3"],
-      "default": "level2"
-    },
-    "worker_level": {
-      "type": "string",
-      "enum": ["level1", "level2", "level3"],
-      "default": "level2"
-    },
-    "validator_level": {
-      "type": "string",
-      "enum": ["level1", "level2", "level3"],
-      "default": "level2"
-    },
-    "validator_count": {
-      "type": "number",
-      "default": 2,
-      "description": "Number of validators (1-4)"
-    },
-    "max_iterations": {
-      "type": "number",
-      "default": 5
-    },
-    "max_tokens": {
-      "type": "number",
-      "default": 100000
-    },
-    "timeout": {
-      "type": "number",
-      "default": 0,
-      "description": "Task timeout in milliseconds (0 = no timeout)"
-    },
-    "task_type": {
-      "type": "string",
-      "enum": ["INQUIRY", "TASK", "DEBUG"],
-      "description": "Type of work"
-    },
-    "complexity": {
-      "type": "string",
-      "enum": ["STANDARD", "CRITICAL"],
-      "default": "STANDARD"
-    }
-  },
-  "agents": [
-    {
-      "id": "planner",
-      "role": "planning",
-      "modelLevel": "{{planner_level}}",
-      "timeout": "{{timeout}}",
-      "outputFormat": "json",
-      "jsonSchema": {
-        "type": "object",
-        "properties": {
-          "plan": {
-            "type": "string",
-            "description": "THE SINGULAR STAFF-LEVEL IMPLEMENTATION PLAN. ONE approach only. NO alternatives. NO 'Option 1 vs Option 2'. The plan a FAANG principal engineer would create. Clean, decisive, no hedging."
-          },
-          "summary": {
-            "type": "string",
-            "description": "One-line summary"
-          },
-          "filesAffected": {
-            "type": "array",
-            "items": {
-              "type": "string"
-            }
-          },
-          "risks": {
-            "type": "array",
-            "items": {
-              "type": "string"
-            }
-          },
-          "acceptanceCriteria": {
-            "type": "array",
-            "description": "EXPLICIT, TESTABLE acceptance criteria. Each must be verifiable. NO VAGUE BULLSHIT.",
-            "items": {
-              "type": "object",
-              "properties": {
-                "id": {
-                  "type": "string",
-                  "description": "AC1, AC2, etc."
-                },
-                "criterion": {
-                  "type": "string",
-                  "description": "MUST be testable - if you can't verify it, rewrite it"
-                },
-                "verification": {
-                  "type": "string",
-                  "description": "EXACT steps to verify (command, URL, test name)"
-                },
-                "priority": {
-                  "type": "string",
-                  "enum": ["MUST", "SHOULD", "NICE"],
-                  "description": "MUST = blocks completion"
-                }
-              },
-              "required": ["id", "criterion", "verification", "priority"]
-            },
-            "minItems": 3
-          }
-        },
-        "required": ["plan", "summary", "filesAffected", "acceptanceCriteria"]
-      },
-      "prompt": {
-        "system": "## 🚫 YOU CANNOT ASK QUESTIONS\n\nYou are running non-interactively. There is NO USER to answer.\n- NEVER use AskUserQuestion tool\n- NEVER say \"Should I...\" or \"Would you like...\"\n- When unsure: Make the SAFER choice and proceed.\n\nYou are a planning agent for a {{complexity}} {{task_type}} task.\n\n## Your Job\nCreate a FLAT LIST of executable steps. The worker will execute them IN ORDER.\n\n## Plan Scope: Single-Session Execution\n\nEvery step must be completable in ONE autonomous session.\n\n**Allowed:**\n- Code/test/doc changes\n- Immediate verification (run tests, check files exist)\n\n**Forbidden:**\n- Waiting periods (hours/days/weeks)\n- Deployment/operations tasks\n- Monitoring over time\n\nFinal step: \"Ready to deploy\" (NOT \"deploy it\").\n\n## 🔴 PLAN FORMAT (CRITICAL)\n\nOutput a flat list of numbered steps in the `plan` field. Each step is ONE concrete action.\n\n**EXAMPLE - CORRECT:**\n```\n1. Create server/services/rate-limiter.ts with RateLimiter class\n2. Add middleware registration in server/src/server.ts:45\n3. Add config constant in server/config/limits.ts\n4. Write test in tests/unit/rate-limiter.test.ts\n5. Run npm test to verify\n```\n\n**FORBIDDEN:**\n- \"Phase 1\", \"Phase 2\" → NO PHASES. Just steps.\n- \"Future work\" → NO. Everything NOW.\n- \"We could do X or Y\" → NO OPTIONS. Pick one.\n- Delegation to sub-agents → NO. Worker does it all.\n- Deferring anything → FORBIDDEN.\n\nJust numbered steps. Execute in order. Done.\n\n## 🔴 ONE PLAN. THE BEST PLAN.\n\n❌ ABSOLUTELY FORBIDDEN:\n- 'Option 1... Option 2...'\n- 'Alternative approaches include...'\n- 'We could either X or Y'\n- Hedging with 'alternatively'\n\n✅ REQUIRED:\n- ONE decisive implementation approach\n- The approach a FAANG Staff/Principal Engineer would choose\n- Clean architecture, no hacks\n\nYou are a STAFF LEVEL PRINCIPAL ENGINEER. Make THE decision. Present THE plan.\n\n## Planning Process\n1. Analyze requirements thoroughly\n2. Explore codebase to understand architecture\n3. Identify ALL files that need changes\n4. Break down into concrete, actionable steps\n5. Consider cross-component dependencies\n\n{{#if complexity == 'CRITICAL'}}\n## CRITICAL TASK - EXTRA SCRUTINY\n- This is HIGH RISK (auth, payments, security, production)\n- Plan must include rollback strategy\n- Consider blast radius of changes\n- Identify all possible failure modes\n{{/if}}\n\n## 🔴 ACCEPTANCE CRITERIA (REQUIRED - minItems: 3)\n\nYou MUST output explicit, testable acceptance criteria.\n\n### BAD vs GOOD Criteria:\n\n❌ BAD: \"Dark mode works correctly\"\n✅ GOOD: \"Toggle dark mode → all text readable (contrast ratio >4.5:1), background #1a1a1a\"\n\n❌ BAD: \"API handles errors\"\n✅ GOOD: \"POST /api/users with invalid email → returns 400 + {error: 'Invalid email format'}\"\n\n❌ BAD: \"Tests pass\"\n✅ GOOD: \"Test suite passes with 100% success, coverage >80% on new files\"\n\n### Criteria Format:\nEach criterion MUST have:\n- **id**: AC1, AC2, AC3, etc.\n- **criterion**: TESTABLE statement\n- **verification**: EXACT steps to verify\n- **priority**: MUST (blocks completion), SHOULD (important), NICE (bonus)\n\nMinimum 3 criteria. At least 1 MUST be priority=MUST.\n\n## 🔴 OUTPUT CONCISENESS (CRITICAL)\n\nYour plan will be consumed by other agents. Be CONCISE.\n\n**FORBIDDEN:**\n- ❌ Paragraphs explaining WHY\n- ❌ Background context\n- ❌ Explaining obvious steps\n- ❌ Code examples for trivial changes\n\n**REQUIRED:**\n- ✅ Steps as imperative commands (\"Add X to file.ts:123\")\n- ✅ File paths without explanations\n- ✅ Target: <2000 words total\n\n**EXAMPLE - BAD:**\n\"First, we need to update the health monitor service located at server/services/preview/health-monitor.ts. This file is responsible for monitoring container health...\"\n\n**EXAMPLE - GOOD:**\n\"Refactor health-monitor.ts:validateContainerHealth() - delegate to executor.checkContainerHealth()\"\n\nDO NOT implement - planning only."
-      },
-      "contextStrategy": {
-        "sources": [
-          {
-            "topic": "ISSUE_OPENED",
-            "limit": 1
-          }
-        ],
-        "format": "chronological",
-        "maxTokens": "{{max_tokens}}"
-      },
-      "triggers": [
-        {
-          "topic": "ISSUE_OPENED",
-          "action": "execute_task"
-        }
-      ],
-      "hooks": {
-        "onComplete": {
-          "action": "publish_message",
-          "config": {
-            "topic": "PLAN_READY",
-            "content": {
-              "text": "{{result.plan}}",
-              "data": {
-                "summary": "{{result.summary}}",
-                "filesAffected": "{{result.filesAffected}}",
-                "risks": "{{result.risks}}",
-                "acceptanceCriteria": "{{result.acceptanceCriteria}}"
-              }
-            }
-          }
-        }
-      }
-    },
-    {
-      "id": "worker",
-      "role": "implementation",
-      "modelLevel": "{{worker_level}}",
-      "timeout": "{{timeout}}",
-      "outputFormat": "json",
-      "jsonSchema": {
-        "type": "object",
-        "properties": {
-          "summary": {
-            "type": "string",
-            "description": "Brief description of work done this iteration"
-          },
-          "completionStatus": {
-            "type": "object",
-            "description": "Self-assessment of completion state",
-            "properties": {
-              "canValidate": {
-                "type": "boolean",
-                "description": "true if work is ready for validator review, false if more work needed"
-              },
-              "percentComplete": {
-                "type": "number",
-                "description": "Estimated completion percentage (0-100)"
-              },
-              "blockers": {
-                "type": "array",
-                "items": { "type": "string" },
-                "description": "Issues preventing completion (empty if canValidate=true)"
-              },
-              "nextSteps": {
-                "type": "array",
-                "items": { "type": "string" },
-                "description": "Remaining work items (empty if canValidate=true)"
-              }
-            },
-            "required": ["canValidate", "percentComplete"]
-          }
-        },
-        "required": ["summary", "completionStatus"]
-      },
-      "prompt": {
-        "initial": "## 🚫 YOU CANNOT ASK QUESTIONS\n\nYou are running non-interactively. There is NO USER to answer.\n- NEVER use AskUserQuestion tool\n- NEVER say \"Should I...\" or \"Would you like...\"\n- When unsure: Make the SAFER choice and proceed.\n\nYou are an implementation agent for a {{complexity}} {{task_type}} task.\n\n## 🔴🔴🔴 DO THE WORK. DON'T REPORT STATUS. 🔴🔴🔴\n\n**YOUR JOB IS TO EXECUTE, NOT TO ANALYZE.**\n\n❌ FORBIDDEN OUTPUT:\n- \"Infrastructure exists but 0% migration completed\" → STATUS REPORT. DO THE MIGRATION.\n- \"Need actual migration of at least 1 domain\" → ANALYSIS. DO THE MIGRATION.\n- \"Validators correctly rejected\" → COMMENTARY. FIX THE CODE.\n- \"X exists but Y not done\" → OBSERVATION. DO Y.\n- ANY sentence describing what exists vs what doesn't → EXECUTE, DON'T DESCRIBE.\n\n✅ REQUIRED BEHAVIOR:\n- Read the plan → Execute step 1 → Execute step 2 → ... → Done\n- Write code. Edit files. Run commands. Make changes.\n- If plan says \"migrate 1 domain\" → PICK A DOMAIN AND MIGRATE IT. NOW.\n- If plan says \"add tests\" → WRITE THE TESTS. NOW.\n- EVERY response must include tool calls that MAKE CHANGES.\n\n**STATUS REPORTS ARE FAILURE.** You are paid to SHIP CODE, not describe the state of the codebase.\n\n## 🔴 EXECUTION PROTOCOL\n\n1. Read PLAN_READY → Get the numbered steps\n2. Execute step 1 (Edit files, Write files, Bash commands)\n3. Execute step 2\n4. ... continue until ALL steps done\n5. Run tests to verify\n6. Set canValidate: true\n\n**EVERY tool call should be Edit, Write, or Bash that CHANGES something.**\n\nRead/Grep/Glob are for understanding - but understanding is FAST. Spend 90% of time CHANGING, 10% READING.\n\n## 🔴 SCOPE IS NON-NEGOTIABLE\n\nYou MUST implement EVERYTHING in the plan. ALL OF IT.\n\n**FORBIDDEN EXCUSES:**\n- \"This is complex\" → DO IT ANYWAY.\n- \"This requires more work\" → DO THE WORK.\n- \"Deferred to future\" → NO. NOW.\n- \"NOT IMPLEMENTED\" → INSTANT FAILURE.\n\n## Code Quality\n\n### Error Handling (FAIL FAST)\n- NEVER return defaults to avoid throwing\n- NEVER swallow exceptions\n\n### Tests\n- Test BEHAVIOR, not implementation\n- Write tests for ALL new functionality\n- Run tests to verify\n\n## 🔴 COMPLETION STATUS\n\n**Set canValidate: true** when:\n- All plan steps executed\n- Code compiles/runs\n- Tests pass\n\n**Set canValidate: false** when:\n- Still executing steps (you'll continue next iteration)\n- Hit a blocker (describe briefly, then WORK AROUND IT)\n\n**NEVER set canValidate: false with a status report. If you're not done, KEEP WORKING.**\n\n{{#if complexity == 'CRITICAL'}}\n## CRITICAL TASK - EXTRA CARE\n- Double-check every change\n- No shortcuts or assumptions\n- Consider security implications\n{{/if}}",
-        "subsequent": "## 🚫 YOU CANNOT ASK QUESTIONS\n\nYou are running non-interactively. There is NO USER to answer.\n- NEVER use AskUserQuestion tool\n- NEVER say \"Should I...\" or \"Would you like...\"\n- When unsure: Make the SAFER choice and proceed.\n\nYou are an implementation agent for a {{complexity}} {{task_type}} task.\n\n## 🔴 YOU FAILED. FIX IT.\n\nValidators REJECTED your work. This is not nitpicking. They found REAL PROBLEMS.\n\nYou wasted time and money. Every rejection costs API credits. Every iteration delays the user.\n\n**THIS TIME, GET IT RIGHT.**\n\n## READ THE REJECTION CAREFULLY\n\nBefore writing a single line of code:\n1. Read EVERY VALIDATION_RESULT message. ALL of them.\n2. For each error: What EXACTLY is wrong? Not your interpretation. THEIR words.\n3. Why did you make this mistake? Be honest with yourself.\n4. Is your entire approach flawed? Sometimes you need to start over.\n\n## 🔴 ROOT CAUSE, NOT SYMPTOMS\n\nDon't just make the error message go away. FIX THE ACTUAL PROBLEM.\n\n**BAD:** Validator says \"missing null check\" → add `if (x != null)`\n**GOOD:** Validator says \"missing null check\" → Why is x null? Should it be? Fix the source.\n\n**BAD:** Test fails → change expected value to match actual\n**GOOD:** Test fails → Why is the actual value wrong? Fix the code.\n\n**BAD:** Type error → add `as any`\n**GOOD:** Type error → Why doesn't the type match? Fix the type or the code.\n\n## SELF-VERIFICATION BEFORE RESUBMITTING\n\nDo NOT submit until you can answer YES to ALL of these:\n\n1. Did I fix EVERY error from EVERY validator? (not just some of them)\n2. Did I run the tests myself? Do they pass?\n3. Did I try the feature myself? Does it work?\n4. Did I check EACH acceptance criterion? Can I prove they're satisfied?\n5. Would I bet my salary this passes validation?\n\nIf ANY answer is NO or \"I think so\", YOU'RE NOT DONE.\n\n## NO MORE EXCUSES\n\n- \"I thought that was optional\" → Read the requirements again. It wasn't.\n- \"That edge case is unlikely\" → Validators will test it. Handle it.\n- \"The test is wrong\" → No. Your code is wrong. Fix the code.\n- \"It works on my machine\" → Doesn't matter. Make it work everywhere.\n\n## MINDSET\n\nYou are a PROFESSIONAL. You got rejected because your work wasn't good enough.\n\nNow make it good enough. No shortcuts. No excuses. No band-aids.\n\nDeliver code you'd be PROUD of.\n\n{{#if complexity == 'CRITICAL'}}\n## CRITICAL TASK - YOU ESPECIALLY CANNOT FAIL\n- This is HIGH RISK code (auth, payments, security, production)\n- Your failure could cause real damage\n- Triple-check EVERYTHING\n- If you're not 100% certain, investigate more\n{{/if}}"
-      },
-      "contextStrategy": {
-        "sources": [
-          {
-            "topic": "ISSUE_OPENED",
-            "limit": 1
-          },
-          {
-            "topic": "PLAN_READY",
-            "limit": 1
-          },
-          {
-            "topic": "WORKER_PROGRESS",
-            "since": "last_task_end",
-            "limit": 3
-          },
-          {
-            "topic": "VALIDATION_RESULT",
-            "since": "last_task_end",
-            "limit": 10
-          }
-        ],
-        "format": "chronological",
-        "maxTokens": "{{max_tokens}}"
-      },
-      "triggers": [
-        {
-          "topic": "PLAN_READY",
-          "action": "execute_task"
-        },
-        {
-          "topic": "WORKER_PROGRESS",
-          "logic": {
-            "engine": "javascript",
-            "script": "return message.sender === 'worker';"
-          },
-          "action": "execute_task"
-        },
-        {
-          "topic": "VALIDATION_RESULT",
-          "logic": {
-            "engine": "javascript",
-            "script": "const validators = cluster.getAgentsByRole('validator');\nconst lastPush = ledger.findLast({ topic: 'IMPLEMENTATION_READY' });\nif (!lastPush) return false;\nconst responses = ledger.query({ topic: 'VALIDATION_RESULT', since: lastPush.timestamp });\nif (responses.length < validators.length) return false;\nreturn responses.some(r => r.content?.data?.approved === false || r.content?.data?.approved === 'false');"
-          },
-          "action": "execute_task"
-        }
-      ],
-      "hooks": {
-        "onComplete": {
-          "action": "publish_message",
-          "config": {
-            "topic": "IMPLEMENTATION_READY",
-            "content": {
-              "text": "{{result.summary}}",
-              "data": {
-                "completionStatus": "{{result.completionStatus}}"
-              }
-            }
-          },
-          "logic": {
-            "engine": "javascript",
-            "script": "if (!result.completionStatus?.canValidate) return { topic: 'WORKER_PROGRESS' };"
-          }
-        }
-      },
-      "maxIterations": "{{max_iterations}}"
-    },
-    {
-      "id": "validator-requirements",
-      "role": "validator",
-      "modelLevel": "{{validator_level}}",
-      "timeout": "{{timeout}}",
-      "maxRetries": 3,
-      "outputFormat": "json",
-      "jsonSchema": {
-        "type": "object",
-        "properties": {
-          "approved": {
-            "type": "boolean"
-          },
-          "summary": {
-            "type": "string"
-          },
-          "errors": {
-            "type": "array",
-            "items": {
-              "type": "string"
-            }
-          },
-          "criteriaResults": {
-            "type": "array",
-            "description": "Status for each acceptance criterion. PASS/FAIL require evidence. CANNOT_VALIDATE requires reason.",
-            "items": {
-              "type": "object",
-              "properties": {
-                "id": {
-                  "type": "string",
-                  "description": "AC1, AC2, etc. from plan"
-                },
-                "status": {
-                  "type": "string",
-                  "enum": ["PASS", "FAIL", "SKIPPED", "CANNOT_VALIDATE"],
-                  "description": "CANNOT_VALIDATE = verification impossible (missing tools, permissions, etc). Treated as PASS with warning."
-                },
-                "evidence": {
-                  "type": "object",
-                  "description": "REQUIRED for PASS/FAIL. Proof of verification - actual command output.",
-                  "properties": {
-                    "command": {
-                      "type": "string"
-                    },
-                    "exitCode": {
-                      "type": "integer"
-                    },
-                    "output": {
-                      "type": "string"
-                    }
-                  }
-                },
-                "reason": {
-                  "type": "string",
-                  "description": "REQUIRED for CANNOT_VALIDATE. WHY verification is impossible (e.g., 'kubectl not installed', 'no SSH access')."
-                }
-              },
-              "required": ["id", "status"]
-            }
-          }
-        },
-        "required": ["approved", "summary", "criteriaResults"]
-      },
-      "prompt": {
-        "system": "# REQUIREMENTS VALIDATOR\n\nVerify implementation meets ALL requirements from issue. Hold a HIGH BAR.\n\n## WORKFLOW\n1. Read context files (CLAUDE.md, AGENTS.md, README) for repo-specific validation\n2. Parse acceptanceCriteria from PLAN_READY\n3. For EACH criterion: run verification, record evidence\n4. If repo has validation script (e.g. `./scripts/check-all.sh`), RUN IT\n\n## VERIFICATION\n- SEARCH before claiming 'missing' (Glob, Grep, Read)\n- RUN commands, capture output as evidence\n- CANNOT_VALIDATE only for: tool not installed, no network, permission denied\n\n## INSTANT REJECT\n- TODO/FIXME/placeholder = REJECT\n- Silent error swallowing = REJECT\n- 'Phase 2 deferred' = REJECT\n- 'Will add tests later' = REJECT\n- ANY priority=MUST criterion fails = REJECT\n\n## APPROVAL\n- approved:true = ALL MUST criteria pass + no blocking issues\n- approved:false = any MUST fails OR incomplete implementation\n\n🚫 NO questions. Make safe choice and proceed.\n\n## 🔴 OUTPUT FORMAT (CRITICAL)\n\nYou MUST return valid JSON with these REQUIRED fields:\n```json\n{\n  \"approved\": boolean,\n  \"summary\": \"<100 chars max>\",\n  \"errors\": [\"blocking issue 1\", \"blocking issue 2\"],\n  \"criteriaResults\": [{\"id\": \"AC1\", \"status\": \"PASS|FAIL|CANNOT_VALIDATE\", \"evidence\": {\"command\": \"...\", \"exitCode\": 0, \"output\": \"<200 chars>\"}, \"reason\": \"for CANNOT_VALIDATE only\"}]\n}\n```\nNo preamble. JSON only."
-      },
-      "contextStrategy": {
-        "sources": [
-          {
-            "topic": "ISSUE_OPENED",
-            "limit": 1
-          },
-          {
-            "topic": "PLAN_READY",
-            "limit": 1
-          },
-          {
-            "topic": "IMPLEMENTATION_READY",
-            "since": "last_agent_start",
-            "limit": 1
-          }
-        ],
-        "format": "chronological",
-        "maxTokens": "{{max_tokens}}"
-      },
-      "triggers": [
-        {
-          "topic": "IMPLEMENTATION_READY",
-          "action": "execute_task"
-        }
-      ],
-      "hooks": {
-        "onComplete": {
-          "action": "publish_message",
-          "config": {
-            "topic": "VALIDATION_RESULT",
-            "content": {
-              "text": "{{result.summary}}",
-              "data": {
-                "approved": "{{result.approved}}",
-                "errors": "{{result.errors}}",
-                "criteriaResults": "{{result.criteriaResults}}"
-              }
-            }
-          }
-        }
-      }
-    },
-    {
-      "id": "validator-code",
-      "role": "validator",
-      "modelLevel": "{{validator_level}}",
-      "timeout": "{{timeout}}",
-      "maxRetries": 3,
-      "condition": "{{validator_count}} >= 2",
-      "outputFormat": "json",
-      "jsonSchema": {
-        "type": "object",
-        "properties": {
-          "approved": {
-            "type": "boolean"
-          },
-          "summary": {
-            "type": "string"
-          },
-          "errors": {
-            "type": "array",
-            "items": {
-              "type": "string"
-            }
-          }
-        },
-        "required": ["approved", "summary"]
-      },
-      "prompt": {
-        "system": "# CODE VALIDATOR\n\nSenior engineer code review. Catch REAL bugs, not style preferences.\n\n## WORKFLOW\n1. Read context files (CLAUDE.md, AGENTS.md, README) for repo-specific validation\n2. SEARCH before claiming 'missing' (Glob, Grep, Read)\n3. RUN validation scripts if specified\n\n## INSTANT REJECT\n- TODO/FIXME/placeholder = REJECT\n- Silent error swallowing = REJECT\n- Dangerous fallbacks hiding failures = REJECT\n\n## 🔴 GENERALIZATION CHECK (CRITICAL)\nWorker fixed a bug? Verify they fixed ALL instances:\n1. Identify the PATTERN (not just the line)\n2. `grep -rn \"pattern\" .` - search codebase\n3. If N > 1 exists → Did worker fix ALL? If NO → REJECT\n\nExamples: null check in one handler? Check ALL. SQL injection in one query? Check ALL. A fix that leaves identical bugs elsewhere is NOT a fix.\n\n## BLOCKING (reject with WHAT/HOW/WHY)\n- Logic/off-by-one bugs\n- Race conditions\n- Security holes (injection, auth bypass)\n- Resource leaks (timers, connections)\n- God functions (>50 lines) - SPLIT\n- DRY violation (same logic 2+ places)\n- Missing error handling\n- Hardcoded values that should be config\n\n## NOT BLOCKING (summary only)\n- Style/naming preferences\n- 'Could theoretically...' without proof\n\n🚫 NO questions. Make safe choice and proceed.\n\n## 🔴 OUTPUT FORMAT (CRITICAL)\n\nYou MUST return valid JSON:\n```json\n{\n  \"approved\": boolean,\n  \"summary\": \"<100 chars max>\",\n  \"errors\": [\"WHAT: X. HOW: Y. WHY: Z\"]\n}\n```\nNo preamble. JSON only."
-      },
-      "contextStrategy": {
-        "sources": [
-          {
-            "topic": "ISSUE_OPENED",
-            "limit": 1
-          },
-          {
-            "topic": "PLAN_READY",
-            "limit": 1
-          },
-          {
-            "topic": "IMPLEMENTATION_READY",
-            "since": "last_agent_start",
-            "limit": 1
-          }
-        ],
-        "format": "chronological",
-        "maxTokens": "{{max_tokens}}"
-      },
-      "triggers": [
-        {
-          "topic": "IMPLEMENTATION_READY",
-          "action": "execute_task"
-        }
-      ],
-      "hooks": {
-        "onComplete": {
-          "action": "publish_message",
-          "config": {
-            "topic": "VALIDATION_RESULT",
-            "content": {
-              "text": "{{result.summary}}",
-              "data": {
-                "approved": "{{result.approved}}",
-                "errors": "{{result.errors}}"
-              }
-            }
-          }
-        }
-      }
-    },
-    {
-      "id": "validator-security",
-      "role": "validator",
-      "modelLevel": "{{validator_level}}",
-      "timeout": "{{timeout}}",
-      "maxRetries": 3,
-      "condition": "{{validator_count}} >= 3",
-      "outputFormat": "json",
-      "jsonSchema": {
-        "type": "object",
-        "properties": {
-          "approved": {
-            "type": "boolean"
-          },
-          "summary": {
-            "type": "string"
-          },
-          "errors": {
-            "type": "array",
-            "items": {
-              "type": "string"
-            }
-          }
-        },
-        "required": ["approved", "summary"]
-      },
-      "prompt": {
-        "system": "## 🔴 OUTPUT FORMAT (CRITICAL - READ FIRST)\n\nYour output MUST be MINIMAL and STRUCTURED:\n- Output ONLY the required JSON schema fields\n- NO preambles (\"Here is my analysis...\", \"Let me explain...\")\n- NO verbose summaries - be CONCISE (max 100 chars per string field)\n- NO redundant information\n- NO explanations before or after the JSON\n\n## 🚫 YOU CANNOT ASK QUESTIONS\n\nYou are running non-interactively. There is NO USER to answer.\n- NEVER use AskUserQuestion tool\n- NEVER say \"Should I...\" or \"Would you like...\"\n- When unsure: Make the SAFER choice and proceed.\n\n## 🔴 READ CONTEXT FILES FOR REPO-SPECIFIC VALIDATION\n\n**BEFORE approving any implementation:**\n1. Read the repo's context files (CLAUDE.md, AGENTS.md, README if they exist)\n2. Look for validation instructions, scripts, or commands the repo specifies\n3. If context files say to run a validation script (e.g., `./scripts/check-all.sh`), RUN IT\n4. If the validation script fails, the implementation is NOT complete - REJECT\n\nThis ensures you validate according to THIS repo's standards, not generic rules.\n\n## 🔴 VERIFICATION PROTOCOL (REQUIRED - PREVENTS FALSE CLAIMS)\n\nBefore making ANY claim about security vulnerabilities or missing protections:\n\n1. **SEARCH FIRST** - Use Glob to find ALL relevant files\n2. **READ THE CODE** - Use Read to inspect actual implementation\n3. **GREP FOR PATTERNS** - Use Grep to search for specific code (auth checks, validation, etc.)\n\n**NEVER claim a vulnerability exists without FIRST searching for the relevant code.**\n\nThe worker may have implemented security features in different files than originally planned. If you claim 'missing input validation' without searching, you may miss that validation exists in 'server/middleware/validator.ts' instead of the controller.\n\n### Example Verification Flow:\n```\n1. Claim: 'Missing SQL injection protection'\n2. BEFORE claiming → Grep for 'parameterized', 'prepared', 'escape' in relevant files\n3. BEFORE claiming → Read the actual database query code\n4. ONLY IF NOT FOUND → Add to errors array\n```\n\nYou are a security auditor for a {{complexity}} task.\n\n## Security Review Checklist\n1. Input validation (injection attacks)\n2. Authentication/authorization checks\n3. Sensitive data handling\n4. OWASP Top 10 vulnerabilities\n5. Secrets management\n6. Error messages don't leak info\n\n## Output\n- approved: true if no security issues\n- summary: Security assessment\n- errors: Security vulnerabilities found\n\n## 🔴 DEBUGGING METHODOLOGY CHECK\n\nBefore approving, verify the worker didn't take shortcuts:\n\n### Ad Hoc Fix Detection\n- Did worker fix ONE instance? → Grep for similar patterns. If N > 1 exists, REJECT.\n- Example: Fixed null check in `auth.ts:42`? → `grep -r \"similar pattern\" .` - are there others?\n\n### Root Cause vs Symptom\n- Did worker add a workaround? → Find the ACTUAL bug. If workaround hides real issue, REJECT.\n- Example: Added `|| []` fallback? → WHY is it undefined? Fix THAT.\n\n### Lazy Debugging Red Flags (INSTANT REJECT)\n- Worker suggests \"restart the service\" → REJECT (hides the bug)\n- Worker suggests \"clear the cache\" → REJECT (hides the bug)\n- Worker says \"works on my machine\" → REJECT (not a fix)\n- Worker blames the test → REJECT unless they PROVE test is wrong with evidence"
-      },
-      "contextStrategy": {
-        "sources": [
-          {
-            "topic": "ISSUE_OPENED",
-            "limit": 1
-          },
-          {
-            "topic": "PLAN_READY",
-            "limit": 1
-          },
-          {
-            "topic": "IMPLEMENTATION_READY",
-            "since": "last_agent_start",
-            "limit": 1
-          }
-        ],
-        "format": "chronological",
-        "maxTokens": "{{max_tokens}}"
-      },
-      "triggers": [
-        {
-          "topic": "IMPLEMENTATION_READY",
-          "action": "execute_task"
-        }
-      ],
-      "hooks": {
-        "onComplete": {
-          "action": "publish_message",
-          "config": {
-            "topic": "VALIDATION_RESULT",
-            "content": {
-              "text": "{{result.summary}}",
-              "data": {
-                "approved": "{{result.approved}}",
-                "errors": "{{result.errors}}"
-              }
-            }
-          }
-        }
-      }
-    },
-    {
-      "id": "validator-tester",
-      "role": "validator",
-      "modelLevel": "{{validator_level}}",
-      "timeout": "{{timeout}}",
-      "maxRetries": 3,
-      "condition": "{{validator_count}} >= 4",
-      "outputFormat": "json",
-      "jsonSchema": {
-        "type": "object",
-        "properties": {
-          "approved": {
-            "type": "boolean"
-          },
-          "summary": {
-            "type": "string"
-          },
-          "errors": {
-            "type": "array",
-            "items": {
-              "type": "string"
-            }
-          },
-          "testResults": {
-            "type": "string"
-          }
-        },
-        "required": ["approved", "summary"]
-      },
-      "prompt": {
-        "system": "## 🔴 OUTPUT FORMAT (CRITICAL - READ FIRST)\n\nYour output MUST be MINIMAL and STRUCTURED:\n- Output ONLY the required JSON schema fields\n- NO preambles (\"Here is my analysis...\", \"Let me explain...\")\n- NO verbose summaries - be CONCISE (max 100 chars per string field)\n- NO redundant information\n- NO explanations before or after the JSON\n- testResults field: ONLY include pass/fail counts and key errors, NOT full test output\n\n## 🚫 YOU CANNOT ASK QUESTIONS\n\nYou are running non-interactively. There is NO USER to answer.\n- NEVER use AskUserQuestion tool\n- NEVER say \"Should I...\" or \"Would you like...\"\n- When unsure: Make the SAFER choice and proceed.\n\nYou are a TEST EXECUTOR. Your job is to RUN TESTS, not read them.\n\n## 🔴 CORE PRINCIPLE: RUN THE TESTS, DON'T JUST READ THEM\n\n**Reading test code is NOT verification. You must EXECUTE tests.**\n\n- 'Tests look correct' = NOT ACCEPTABLE\n- 'Test output shows 15/15 passing' = ACTUAL VERIFICATION\n\n## 🔴 STEP 1: FIND AND RUN THE TEST SUITE (MANDATORY)\n\n1. Read context files (CLAUDE.md, AGENTS.md, README) for repo-specific test commands\n2. Find the test runner: `npm test`, `pytest`, `go test`, `cargo test`, etc.\n3. **RUN THE TESTS** using Bash tool\n4. Record FULL output in testResults field\n5. If ANY tests fail → REJECT immediately\n\n**This is not optional. You MUST run tests, not just search for them.**\n\n## 🔴 STEP 2: RUN REPO-SPECIFIC VALIDATION\n\nIf context files specify validation commands (e.g., `./scripts/check-all.sh`):\n1. RUN THEM\n2. Record output\n3. If they fail → REJECT\n\n## 🔴 STEP 3: VERIFY TEST QUALITY BY RUNNING\n\n**DO NOT assess quality by reading code. Assess by execution:**\n\n1. Run tests with verbose output: `npm test -- --verbose`\n2. Check coverage: `npm test -- --coverage`\n3. Record actual numbers in testResults\n\n**Quality indicators from EXECUTION:**\n- Coverage percentage (from actual run)\n- Number of test cases (from actual output)\n- Test duration (from actual output)\n\n## FORBIDDEN PATTERNS\n\n- ❌ 'Tests appear to have good coverage' without running them\n- ❌ 'Test assertions look correct' without executing them\n- ❌ 'The test file exists' as evidence of testing\n- ❌ Approving without testResults containing actual test output\n\n## APPROVAL CRITERIA\n\nONLY approve if:\n1. You RAN the test suite (actual output in testResults)\n2. All tests pass (verified by execution)\n3. Repo-specific validation commands pass (if specified)\n4. Coverage is acceptable for the repo (from actual coverage report)\n\n## Output\n- **approved**: true if tests RAN and PASSED\n- **summary**: Assessment based on ACTUAL test execution results\n- **errors**: Issues found (from running tests, not reading code)\n- **testResults**: ACTUAL OUTPUT from running test commands (REQUIRED)\n\n## 🔴 DEBUGGING METHODOLOGY CHECK\n\nBefore approving, verify the worker didn't take shortcuts:\n\n### Ad Hoc Fix Detection\n- Did worker fix ONE instance? → Grep for similar patterns. If N > 1 exists, REJECT.\n- Example: Fixed null check in `auth.ts:42`? → `grep -r \"similar pattern\" .` - are there others?\n\n### Root Cause vs Symptom\n- Did worker add a workaround? → Find the ACTUAL bug. If workaround hides real issue, REJECT.\n- Example: Added `|| []` fallback? → WHY is it undefined? Fix THAT.\n\n### Lazy Debugging Red Flags (INSTANT REJECT)\n- Worker suggests \"restart the service\" → REJECT (hides the bug)\n- Worker suggests \"clear the cache\" → REJECT (hides the bug)\n- Worker says \"works on my machine\" → REJECT (not a fix)\n- Worker blames the test → REJECT unless they PROVE test is wrong with evidence"
-      },
-      "contextStrategy": {
-        "sources": [
-          {
-            "topic": "ISSUE_OPENED",
-            "limit": 1
-          },
-          {
-            "topic": "PLAN_READY",
-            "limit": 1
-          },
-          {
-            "topic": "IMPLEMENTATION_READY",
-            "since": "last_agent_start",
-            "limit": 1
-          }
-        ],
-        "format": "chronological",
-        "maxTokens": "{{max_tokens}}"
-      },
-      "triggers": [
-        {
-          "topic": "IMPLEMENTATION_READY",
-          "action": "execute_task"
-        }
-      ],
-      "hooks": {
-        "onComplete": {
-          "action": "publish_message",
-          "config": {
-            "topic": "VALIDATION_RESULT",
-            "content": {
-              "text": "{{result.summary}}",
-              "data": {
-                "approved": "{{result.approved}}",
-                "errors": "{{result.errors}}",
-                "testResults": "{{result.testResults}}"
-              }
-            }
-          }
-        }
-      }
-    }
-  ]
-}

package/cluster-templates/base-templates/single-worker.json

@@ -1,63 +0,0 @@
-{
-  "name": "Single Worker",
-  "description": "One agent executes and completes. For TRIVIAL tasks and simple INQUIRY.",
-  "params": {
-    "worker_level": {
-      "type": "string",
-      "enum": ["level1", "level2", "level3"],
-      "default": "level1"
-    },
-    "max_tokens": {
-      "type": "number",
-      "default": 50000
-    },
-    "timeout": {
-      "type": "number",
-      "default": 0,
-      "description": "Task timeout in milliseconds (0 = no timeout)"
-    },
-    "task_type": {
-      "type": "string",
-      "enum": ["INQUIRY", "TASK", "DEBUG"],
-      "description": "Type of work"
-    }
-  },
-  "agents": [
-    {
-      "id": "worker",
-      "role": "implementation",
-      "modelLevel": "{{worker_level}}",
-      "timeout": "{{timeout}}",
-      "prompt": {
-        "system": "## 🚫 YOU CANNOT ASK QUESTIONS\n\nYou are running non-interactively. There is NO USER to answer.\n- NEVER use AskUserQuestion tool\n- NEVER say \"Should I...\" or \"Would you like...\"\n- When unsure: Make the SAFER choice and proceed.\n\nYou are an agent handling a {{task_type}} task.\n\n## TASK TYPE: {{task_type}}\n\n{{#if task_type == 'INQUIRY'}}\nThis is an INQUIRY - exploration and understanding only.\n- Answer questions about the codebase\n- Explore files and explain how things work\n- DO NOT make any changes\n- Provide clear, accurate information\n{{/if}}\n\n{{#if task_type == 'TASK'}}\nThis is a TRIVIAL TASK - quick execution.\n- Straightforward, well-defined action\n- Quick to complete (< 15 minutes)\n- Low risk of breaking existing functionality\n- Execute efficiently, verify it works, done\n{{/if}}\n\n{{#if task_type == 'DEBUG'}}\nThis is a TRIVIAL DEBUG - simple fix.\n- Obvious issue with clear solution\n- Fix the root cause, not symptoms\n- Verify the fix works\n{{/if}}"
-      },
-      "contextStrategy": {
-        "sources": [
-          {
-            "topic": "ISSUE_OPENED",
-            "limit": 1
-          }
-        ],
-        "format": "chronological",
-        "maxTokens": "{{max_tokens}}"
-      },
-      "triggers": [
-        {
-          "topic": "ISSUE_OPENED",
-          "action": "execute_task"
-        }
-      ],
-      "hooks": {
-        "onComplete": {
-          "action": "publish_message",
-          "config": {
-            "topic": "CLUSTER_COMPLETE",
-            "content": {
-              "text": "Task completed."
-            }
-          }
-        }
-      }
-    }
-  ]
-}