principles-disciple 1.8.2 → 1.8.3

package/templates/langs/en/skills/ai-sprint-orchestration/references/specs/feature-complex-template.json

@@ -0,0 +1,107 @@
+{
+  "id": "feature-complex-template",
+  "title": "Complex Feature Template - Replace before use",
+  "workspace": "__SKILL_PACKAGE_ROOT__",
+  "branchWorkspace": "__SKILL_PACKAGE_ROOT__",
+  "branch": "main",
+  "requiresTaskContract": true,
+  "taskContract": {
+    "goal": "<describe the feature to implement and the user-visible behavior>",
+    "inScope": [
+      "<allowed modules or files>"
+    ],
+    "outOfScope": [
+      "<forbidden scope, unrelated refactors, external product-side work>"
+    ],
+    "validationCommands": [
+      "<command 1>",
+      "<command 2>"
+    ],
+    "expectedArtifacts": [
+      "decision.md",
+      "scorecard.json",
+      "<optional feature artifact>"
+    ]
+  },
+  "executionScope": {
+    "maxFiles": 8,
+    "maxChecks": 5,
+    "maxDeliverables": 4
+  },
+  "maxRoundsPerStage": 2,
+  "maxRuntimeMinutes": 75,
+  "stageTimeoutMinutes": 20,
+  "stages": ["investigate", "implement-pass-1", "verify"],
+  "producer": {
+    "agent": "iflow",
+    "model": "glm-4.7",
+    "timeoutSeconds": 240
+  },
+  "reviewerA": {
+    "agent": "iflow",
+    "model": "glm-4.7",
+    "timeoutSeconds": 120,
+    "focus": "Validate feature behavior, acceptance criteria coverage, and architecture fit."
+  },
+  "reviewerB": {
+    "agent": "iflow",
+    "model": "glm-4.7",
+    "timeoutSeconds": 120,
+    "focus": "Validate implementation scope, verification evidence, and regression risk."
+  },
+  "context": [
+    "This is a generic complex feature sprint template.",
+    "Replace all placeholder contract values before running.",
+    "Use it for bounded feature delivery with explicit acceptance criteria and verification commands."
+  ],
+  "stageGoals": {
+    "investigate": [
+      "Clarify feature scope and constraints.",
+      "Identify the smallest valuable slice that can be implemented now.",
+      "Produce a verification plan aligned with acceptance criteria."
+    ],
+    "implement-pass-1": [
+      "Implement the smallest valuable feature slice within the declared scope.",
+      "Run the planned verification commands.",
+      "Record evidence that the new behavior exists and is bounded."
+    ],
+    "verify": [
+      "Verify the feature behaves as intended.",
+      "Check whether the acceptance criteria are actually covered by evidence.",
+      "Decide whether the feature is ready, needs continuation, or revise."
+    ]
+  },
+  "stageCriteria": {
+    "investigate": {
+      "requiredApprovals": 2,
+      "requiredProducerSections": ["SUMMARY", "CHANGES", "EVIDENCE", "CODE_EVIDENCE", "KEY_EVENTS", "HYPOTHESIS_MATRIX", "CHECKS", "OPEN_RISKS"],
+      "requiredReviewerSections": ["VERDICT", "BLOCKERS", "FINDINGS", "CODE_EVIDENCE", "HYPOTHESIS_MATRIX", "NEXT_FOCUS", "CHECKS"],
+      "scoringDimensions": ["scope_definition", "acceptance_alignment"],
+      "dimensionThreshold": 3,
+      "requiredDeliverables": ["feature_slice", "acceptance_plan", "verification_plan"]
+    },
+    "implement-pass-1": {
+      "requiredApprovals": 2,
+      "requiredProducerSections": ["SUMMARY", "CHANGES", "EVIDENCE", "CODE_EVIDENCE", "KEY_EVENTS", "HYPOTHESIS_MATRIX", "CHECKS", "OPEN_RISKS", "CONTRACT"],
+      "requiredReviewerSections": ["VERDICT", "BLOCKERS", "FINDINGS", "CODE_EVIDENCE", "HYPOTHESIS_MATRIX", "NEXT_FOCUS", "CHECKS"],
+      "scoringDimensions": ["behavior_delivery", "verification_quality", "scope_control"],
+      "dimensionThreshold": 3,
+      "requiredDeliverables": ["feature_implementation", "evidence_of_behavior", "risk_summary"]
+    },
+    "verify": {
+      "requiredApprovals": 2,
+      "requiredProducerSections": ["SUMMARY", "CHANGES", "EVIDENCE", "CODE_EVIDENCE", "KEY_EVENTS", "HYPOTHESIS_MATRIX", "CHECKS", "OPEN_RISKS"],
+      "requiredReviewerSections": ["VERDICT", "BLOCKERS", "FINDINGS", "CODE_EVIDENCE", "HYPOTHESIS_MATRIX", "NEXT_FOCUS", "CHECKS"],
+      "scoringDimensions": ["artifact_verification", "acceptance_coverage"],
+      "dimensionThreshold": 3,
+      "requiredDeliverables": ["acceptance_verified", "residual_risk_assessment"]
+    }
+  },
+  "stageRoleTimeouts": {
+    "investigate": { "producer": 180, "reviewer_a": 120, "reviewer_b": 120 },
+    "implement-pass-1": { "producer": 300, "reviewer_a": 120, "reviewer_b": 120 },
+    "verify": { "producer": 120, "reviewer_a": 90, "reviewer_b": 90 }
+  },
+  "continuationSpec": null,
+  "recoverySpec": null
+}

package/templates/langs/en/skills/ai-sprint-orchestration/references/specs/workflow-validation-minimal-verify.json

@@ -0,0 +1,105 @@
+{
+  "id": "workflow-validation-minimal-verify",
+  "title": "Verify: Workflow v1 infrastructure validation",
+  "workspace": "__SKILL_PACKAGE_ROOT__",
+  "branchWorkspace": "__SKILL_PACKAGE_ROOT__",
+  "branch": "main",
+  "integrationPhase": "shadow",
+  "maxRoundsPerStage": 1,
+  "maxRuntimeMinutes": 15,
+  "stageTimeoutMinutes": 5,
+  "stages": ["verify"],
+  "producer": {
+    "agent": "iflow",
+    "model": "glm-4.7",
+    "timeoutSeconds": 90
+  },
+  "reviewerA": {
+    "agent": "opencode",
+    "model": "minimax-cn-coding-plan/MiniMax-M2.7",
+    "timeoutSeconds": 45,
+    "fallback": {
+      "agent": "iflow",
+      "model": "glm-4.7",
+      "timeoutSeconds": 75
+    },
+    "role": "artifact_checker",
+    "focus": "Verify decision.md and scorecard.json contain expected fields"
+  },
+  "reviewerB": {
+    "agent": "opencode",
+    "model": "minimax-cn-coding-plan/MiniMax-M2.7",
+    "timeoutSeconds": 45,
+    "fallback": {
+      "agent": "iflow",
+      "model": "glm-4.7",
+      "timeoutSeconds": 75
+    },
+    "role": "recommendation_checker",
+    "focus": "Verify nextRunRecommendation is computed and reasonable"
+  },
+  "context": [
+    "=== VERIFY RUN ===",
+    "This run verifies the output of workflow-validation-minimal sprint.",
+    "",
+    "=== VERIFICATION CHECKLIST ===",
+    "1. Check sprint directory exists under runtime/runs/<run-id>/",
+    "2. Check decision.md exists and contains:",
+    "   - Outcome: (advance|revise|halt)",
+    "   - outputQuality: (shadow_complete|production_ready|needs_work)",
+    "   - qualityReasons: (list of strings)",
+    "3. Check scorecard.json exists and contains:",
+    "   - outputQuality field",
+    "   - qualityReasons array",
+    "   - validation object with errorSummary",
+    "4. Check nextRunRecommendation is recorded if outputQuality != production_ready",
+    "",
+    "=== PRODUCER TASK ===",
+    "Read the sprint directory from the previous run and report findings.",
+    "Use the run-id provided in the stage brief."
+  ],
+  "stageGoals": {
+    "verify": [
+      "Verify previous run produced valid artifacts.",
+      "Confirm outputQuality was persisted correctly.",
+      "Confirm nextRunRecommendation was computed."
+    ]
+  },
+  "workUnits": {
+    "verify": [
+      {
+        "workUnitId": "verify-artifact-slice",
+        "workUnitGoal": "Verify the prior validation run using compact artifact-focused context.",
+        "allowedFiles": [
+          "runtime/runs",
+          "scripts/run.mjs",
+          "references/specs/workflow-validation-minimal-verify.json"
+        ],
+        "unitChecks": [
+          "node scripts/run.mjs --task workflow-validation-minimal"
+        ],
+        "unitDeliverables": [
+          "artifacts verified",
+          "fields checked"
+        ],
+        "unitSummary": "Verify persisted workflow artifacts without reopening broad stage history.",
+        "carryForwardSummary": "Use prior run id and compact artifact checklist first; avoid long historical replay."
+      }
+    ]
+  },
+  "stageCriteria": {
+    "verify": {
+      "requiredApprovals": 2,
+      "requiredProducerSections": ["SUMMARY", "EVIDENCE", "CODE_EVIDENCE", "CHECKS"],
+      "requiredReviewerSections": ["VERDICT", "BLOCKERS", "FINDINGS", "CODE_EVIDENCE", "NEXT_FOCUS", "CHECKS"],
+      "scoringDimensions": ["artifact_verification", "field_correctness"],
+      "dimensionThreshold": 3,
+      "requiredDeliverables": ["artifacts_verified", "fields_correct"]
+    }
+  },
+  "stageRoleTimeouts": {
+    "verify": { "producer": 90, "reviewer_a": 45, "reviewer_b": 45 }
+  },
+  "continuationSpec": null,
+  "recoverySpec": null
+}

package/templates/langs/en/skills/ai-sprint-orchestration/references/specs/workflow-validation-minimal.json

@@ -0,0 +1,108 @@
+{
+  "id": "workflow-validation-minimal",
+  "title": "Workflow v1 Minimal Validation - Test workflow infrastructure without product binding",
+  "workspace": "__SKILL_PACKAGE_ROOT__",
+  "branchWorkspace": "__SKILL_PACKAGE_ROOT__",
+  "branch": "main",
+  "integrationPhase": "shadow",
+  "maxRoundsPerStage": 2,
+  "maxRuntimeMinutes": 20,
+  "stageTimeoutMinutes": 4,
+  "stages": ["validate"],
+  "producer": {
+    "agent": "iflow",
+    "model": "glm-4.7",
+    "timeoutSeconds": 90
+  },
+  "reviewerA": {
+    "agent": "opencode",
+    "model": "minimax-cn-coding-plan/MiniMax-M2.7",
+    "timeoutSeconds": 45,
+    "fallback": {
+      "agent": "iflow",
+      "model": "glm-4.7",
+      "timeoutSeconds": 75
+    },
+    "role": "schema_checker",
+    "focus": "Verify report has all required sections, VERDICT format, DIMENSIONS format"
+  },
+  "reviewerB": {
+    "agent": "opencode",
+    "model": "minimax-cn-coding-plan/MiniMax-M2.7",
+    "timeoutSeconds": 45,
+    "fallback": {
+      "agent": "iflow",
+      "model": "glm-4.7",
+      "timeoutSeconds": 75
+    },
+    "role": "quality_checker",
+    "focus": "Verify outputQuality is computed correctly, qualityReasons are meaningful"
+  },
+  "context": [
+    "=== WORKFLOW VALIDATION SPRINT ===",
+    "This sprint validates the workflow infrastructure itself, NOT any product feature.",
+    "",
+    "=== VALIDATION GOALS ===",
+    "1. Producer produces a valid report with all required sections",
+    "2. Reviewers produce valid reports with VERDICT, DIMENSIONS",
+    "3. Orchestrator computes outputQuality correctly",
+    "4. outputQuality and qualityReasons are persisted to decision.md and scorecard.json",
+    "5. nextRunRecommendation is computed and recorded",
+    "",
+    "=== NO PRODUCT LOGIC ===",
+    "Producer task: List 3 files in packages/openclaw-plugin/templates/langs/zh/skills/ai-sprint-orchestration/scripts/ and describe their purpose.",
+    "This is intentionally trivial - we are testing the workflow package, not any product task.",
+    "",
+    "=== SUCCESS CRITERIA FOR THIS RUN ===",
+    "- No halt due to workflow infrastructure issues (timeout handling, file I/O, validation)",
+    "- decision.md contains outputQuality field",
+    "- scorecard.json contains outputQuality and qualityReasons",
+    "- If outputQuality is shadow_complete, nextRunRecommendation suggests CONTINUATION or VERIFY"
+  ],
+  "stageGoals": {
+    "validate": [
+      "Produce a minimal valid producer report.",
+      "Reviewers verify schema compliance.",
+      "Orchestrator computes and persists outputQuality."
+    ]
+  },
+  "workUnits": {
+    "validate": [
+      {
+        "workUnitId": "validate-contract-slice",
+        "workUnitGoal": "Validate the minimal workflow package using a single compact work unit.",
+        "allowedFiles": [
+          "scripts/run.mjs",
+          "scripts/lib/task-specs.mjs",
+          "references/specs/workflow-validation-minimal.json"
+        ],
+        "unitChecks": [
+          "node scripts/run.mjs --self-check"
+        ],
+        "unitDeliverables": [
+          "producer report drafted",
+          "self-check executed"
+        ],
+        "unitSummary": "Use one small work unit to validate package workflow behavior.",
+        "carryForwardSummary": "Prefer compact workflow-only context. Do not expand into product-side closure."
+      }
+    ]
+  },
+  "stageCriteria": {
+    "validate": {
+      "requiredApprovals": 2,
+      "requiredProducerSections": ["SUMMARY", "CHANGES", "EVIDENCE", "CODE_EVIDENCE", "KEY_EVENTS", "HYPOTHESIS_MATRIX", "CHECKS", "OPEN_RISKS"],
+      "requiredReviewerSections": ["VERDICT", "BLOCKERS", "FINDINGS", "CODE_EVIDENCE", "HYPOTHESIS_MATRIX", "NEXT_FOCUS", "CHECKS"],
+      "scoringDimensions": ["schema_compliance", "report_completeness"],
+      "dimensionThreshold": 3,
+      "requiredDeliverables": ["producer_report_complete", "self_check_passed"]
+    }
+  },
+  "stageRoleTimeouts": {
+    "validate": { "producer": 60, "reviewer_a": 45, "reviewer_b": 45 }
+  },
+  "verificationSpec": "workflow-validation-minimal-verify",
+  "continuationSpec": null,
+  "recoverySpec": null,
+  "requireVerify": false
+}

package/templates/langs/en/skills/ai-sprint-orchestration/references/workflow-v1-acceptance-checklist.md

@@ -0,0 +1,58 @@
+# Workflow v1 Acceptance Checklist
+
+## Purpose
+
+Use this checklist to validate that the packaged workflow is readable, executable, and safe to hand off to another agent.
+
+## Acceptance gates
+
+- [ ] Baseline tests are green
+- [ ] `workflow-validation-minimal` completes and writes package-local artifacts
+- [ ] `workflow-validation-minimal-verify` completes and validates the previous run
+- [ ] Every failure is classified into the approved four-category taxonomy
+
+## Commands
+
+```powershell
+node scripts/run.mjs --self-check
+node scripts/run.mjs --help
+node scripts/run.mjs --task workflow-validation-minimal
+node scripts/run.mjs --task workflow-validation-minimal-verify
+```
+
+## Run result record
+
+| Field | Value |
+|------|------|
+| runId | |
+| outcome | |
+| outputQuality | |
+| validation | |
+| nextRunRecommendation | |
+| failure classification | |
+
+## Required artifact checks
+
+- `decision.md` contains `outputQuality`
+- `decision.md` contains `qualityReasons`
+- `decision.md` contains readable validation status
+- `scorecard.json` contains `outputQuality`
+- `scorecard.json` contains `qualityReasons`
+- `scorecard.json` contains `validation`
+- `scorecard.json` contains `nextRunRecommendation`
+
+## Failure classification
+
+- `workflow bug`
+- `agent behavior issue`
+- `environment issue`
+- `sample-spec issue`
+
+## Stop conditions
+
+If the run exposes a sample-side or product-side issue:
+
+- classify it
+- record evidence
+- stop the run review
+- do not continue into product closure work

package/templates/langs/en/skills/ai-sprint-orchestration/references/workflow-v1.4-work-unit-handoff.md

@@ -0,0 +1,190 @@
+# Workflow v1.4 Work-Unit Upgrade Handoff
+
+## Goal
+
+Use the packaged `ai-sprint-orchestration` skill to implement the next architecture upgrade of the workflow itself.
+
+This upgrade is **not** about product-side closure. It is about making the orchestrator better at long-running, high-value coding tasks by introducing a finer-grained execution model.
+
+The target direction is:
+
+- move from `stage -> producer/reviewer`
+- toward `stage -> work units -> producer/reviewer/decision`
+
+The key design goal is:
+
+- smaller execution units
+- explicit checkpointing
+- forced context reload between units
+- less long-context drift
+
+## Current confirmed state
+
+The packaged skill is already usable as an internal tool:
+
+- package path:
+  - `D:/Code/principles/packages/openclaw-plugin/templates/langs/zh/skills/ai-sprint-orchestration`
+- installed path:
+  - `C:/Users/Administrator/.agents/skills/ai-sprint-orchestration`
+- self-check works
+- package-local validation runs work
+- failure classification is persisted to:
+  - `latest-summary.md`
+  - `scorecard.json`
+- complex task templates and minimum task contract already exist
+
+Remaining instability is mostly:
+
+- `agent behavior issue`
+
+It is **not** primarily:
+
+- pathing
+- packaging
+- runtime-root layout
+- missing acceptance artifacts
+
+## Why v1.4 is needed
+
+The current workflow already externalizes state and can resume from artifacts, but its granularity is still too coarse for very complex tasks.
+
+Today it mainly refreshes context at:
+
+- stage level
+- role level
+- round level
+
+That is useful, but not yet enough for difficult long-running work where one producer pass can still become too large, too noisy, or too drift-prone.
+
+The next upgrade should add a smaller unit below stage:
+
+- `workUnitId`
+- `workUnitGoal`
+- `allowedFiles`
+- `unitChecks`
+- `unitDeliverables`
+- `unitSummary`
+- `carryForwardSummary`
+
+## Scope
+
+### In scope
+
+- `packages/openclaw-plugin/templates/langs/zh/skills/ai-sprint-orchestration`
+- `packages/openclaw-plugin/templates/langs/en/skills/ai-sprint-orchestration`
+- workflow docs, specs, prompts, state shape, and summary artifacts
+- package-local validation and workflow-only tests
+
+### Out of scope
+
+- `packages/openclaw-plugin`
+- `D:/Code/openclaw`
+- product-side/sample-side closure
+- dashboard
+- stageGraph
+- self-optimizing sprint
+- parallel orchestrator expansion
+
+If validation or implementation exposes product-side or sample-side issues, classify them and stop. Do not drift back into product closure work.
+
+## Architectural hypothesis
+
+The most valuable next step is **not** a full rewrite.
+
+The most valuable next step is to implement a high-value foundation slice:
+
+1. define the work-unit contract
+2. define how a stage declares work units
+3. add package-local templates and docs for work-unit-aware tasks
+4. add checkpoint/carry-forward artifacts that are short enough to be reused safely
+5. update prompts so each unit starts from minimal context, not long historical text
+
+This means the first v1.4 sprint should prioritize:
+
+- work-unit interfaces
+- unit-level artifacts
+- unit-level carry-forward
+- validation of the new context-reload behavior
+
+It should **not** try to finish every future v1.4 feature in one pass.
+
+## Recommended implementation order
+
+### Phase 1: foundation interfaces
+
+Define and document the unit contract in specs and references:
+
+- `workUnitId`
+- `workUnitGoal`
+- `allowedFiles`
+- `unitChecks`
+- `unitDeliverables`
+- `unitSummary`
+- `carryForwardSummary`
+
+Add minimal schema/contract validation for these fields where appropriate.
+
+### Phase 2: unit-scoped prompt inputs
+
+Update producer/reviewer prompt construction so a unit run receives:
+
+- current unit goal
+- allowed files
+- expected checks
+- expected deliverables
+- prior carry-forward summary
+
+Prefer short checkpoint summaries over replaying long decision history.
+
+### Phase 3: checkpoint and continuation tightening
+
+Standardize the artifact that one unit leaves behind for the next:
+
+- accomplished
+- blockers
+- next focus
+- verified files
+
+Make continuation/revise paths prefer this short artifact.
+
+### Phase 4: validation
+
+Add or update validation specs and tests so the workflow can prove:
+
+- missing task contract is rejected
+- work-unit metadata is loaded correctly
+- continuation reads compact carry-forward context
+- failures still classify cleanly
+
+## Success criteria
+
+The first v1.4 slice is successful if all of these are true:
+
+1. The workflow can express work-unit metadata in package-local specs or templates.
+2. Prompts clearly scope a run to a smaller work unit, not a broad stage-only brief.
+3. Continuation uses compact carry-forward summaries by default.
+4. Baseline tests pass.
+5. Package-local self-check still passes.
+6. Validation runs either pass, or fail with explicit classification that is not a workflow-plumbing ambiguity.
+
+## Guardrails
+
+- Do not rewrite the orchestrator from scratch.
+- Do not introduce a second orchestration system.
+- Do not move back into product/sample closure.
+- Do not increase scope to unrelated features.
+- Prefer one workflow-only architectural improvement per iteration.
+- Preserve the packaged skill as the primary operator entry point.
+
+## Suggested starting point for the next thread
+
+In the next thread, the first sprint should focus on:
+
+- implementing the work-unit contract and carry-forward foundation
+- not the entire future v1.4 roadmap
+
+Treat this as a bounded architecture slice:
+
+- high value
+- low drift
+- directly useful for future complex coding tasks

package/templates/langs/en/skills/ai-sprint-orchestration/runtime/.gitignore

@@ -0,0 +1,2 @@
+*
+!.gitignore