{"id":"implement-plan-micro","name":"implement_plan_micro","summary":"考えや共有、計画から技術計画を実行し、検証を行ってください","body":"# Formal Specification\n\n## Multimodal Logic Integration\n\nFive modal logics via fusion with bridge principles:\n- **JL**: Justification Logic - evidence-backed claims\n- **IEL**: Inferential Erotetic Logic - question handling\n- **TEL**: Temporal Epistemic Logic - phase sequencing\n- **SDL**: Standard Deontic Logic - obligations/permissions\n- **DEL**: Dynamic Epistemic Logic - action modalities\n\n## Justification Logic (JL)\n\n```\n# Justification terms\n[h]:context(task_n) # Handoff h justifies task context\n[v]:verified(phase_n) # Verification v justifies completion\n[p]:plan(tasks) # Plan p justifies task list\n\n# Evidence production\n[read(f)]exists e. [e]:content(f)\n[verify(c)]exists v. [v]:pass(c) | [v]:fail(c)\n\n# Handoff chain: evidence propagates\n[h_n]:complete(task_n) -> [h_{n+1}]:context(task_{n+1})\nproceed(task) <-> exists h. [h]:validated\n```\n\n## Inferential Erotetic Logic (IEL)\n\n```\n# Mode and blocker questions\n?{direct, orchestration} # Mode selection\n?{continue, retry, ask_user} # Blocker resolution\nmismatch(plan, reality) -> ?{how_proceed}\nno_validation -> ?{run_validation_first}\n```\n\n## Temporal Epistemic Logic (TEL)\n\n```\n# File reading constraints\n[](mentioned(f) -> <>read_fully(f)) # Eventually read\n[](mentioned(f) -> not spawn U read_fully(f)) # No spawn until read\n[](partial_read(f) -> false) # Partial reads forbidden\n\n# Phase sequencing\n[](phase(n) -> P(phase(n-1) & verified(n-1))) # Verified before next\n[](automated_pass -> <>manual_verify) # Automated gates manual\n[](manual_pass(n) -> <>phase(n+1)) # Manual gates next phase\n\n# Handoff persistence\n[](handoff_created(h) -> []exists_on_disk(h)) # Survives compaction\n\n# Termination\n<>(all_complete | abandoned)\n```\n\n## Standard Deontic Logic (SDL)\n\n```\n# Reading obligations\nO(read_fully(plan))\nO(read_fully(f)) <- mentioned_in_plan(f)\nO(check_existing_checkmarks)\nF(partial_read)\n\n# Verification obligations\nO(run_automated) <- impl_complete\nO(pause_for_manual) <- automated_pass\nO(present_manual_checklist)\nF(checkoff_manual) <- not user_confirmed\n\n# Mode selection\nO(orchestration) <- tasks >= 4\nP(direct) <- tasks <= 3\nO(respect_user_preference)\n\n# Orchestration obligations\nO(read_previous_handoff) <- exists_handoff(task_{n-1})\nO(create_handoff) <- agent_completes\nO(update_ledger) <- task_complete\nF(batch_tasks) # One agent per task\nF(proceed_on_mismatch) <- not user_guidance\n```\n\n## Dynamic Epistemic Logic (DEL)\n\n```\n# Implementation actions\n[read(plan)]K(tasks) & K(phases) & K(criteria)\n[read(handoff_n)]K(context_{n+1})\n[spawn(agent, task)]<>result(agent)\n[verify(c)](K(pass) | K(fail))\n\n# Composed workflows\n[select_direct][implement ; verify_auto ; present_manual ; wait]*\n[select_orchestration][prepare ; spawn ; wait ; read_handoff ; update]*\n\n# Recovery\n[compaction ; read_ledger ; list_handoffs ; read_last]resume\n\n# Mismatch\n[detect_mismatch ; stop ; present ; wait]proceed_or_abort\n```\n\n## Bridge Principles\n\n```\n# Evidence persistence (JL-TEL)\n[h]:context(n) -> [][h]:context(n)\n\n# Evidence obligations (JL-SDL)\nO(exists h. [h]:validated) <- pre_implement\nO(exists v. [v]:pass(auto)) <- pre_manual\n\n# Handoff chain (full integration)\n[h_n]:complete(n) -> O([spawn]<>[h_{n+1}]:context(n+1))\ncompaction -> (forall h. persists(h))\n```\n\n## State Machine\n\n```\nINIT --> READ_PLAN --> MODE_SELECT --+--> DIRECT: [IMPL -> AUTO -> MANUAL -> WAIT]*\n |\n +--> ORCHESTRATION: [PREP -> SPAWN -> WAIT -> HANDOFF]*\n |\n v\n COMPLETE\n```\n\n## Output Schema\n\n```yaml\nhandoff_path: \"thoughts/handoffs//task-[NN]-[desc].md\"\nschema:\n required: [status, task_desc, files_modified[], verification_results, context_for_next]\n optional: [blocker, decisions[], open_questions[]]\ntracking:\n plan: \"- [x] Task N: description\"\n ledger: \"[x] Task N\"\n```\n\n---\n\n# Prose (Where Logic Insufficient)\n\n## Mode Selection\n\n| Tasks | Context Critical | Mode |\n|-------|------------------|------|\n| 1-3 | No | Direct |\n| 1-3 | Yes | Orchestration |\n| 4+ | Any | Orchestration |\n\nUser preference overrides.\n\n## Templates\n\n**Mismatch:**\n```\nIssue in Phase [N]:\nExpected: [plan says]\nFound: [actual]\nHow should I proceed?\n```\n\n**Manual Verification Pause:**\n```\nPhase [N] Complete - Ready for Manual Verification\nAutomated passed: [list]\nPlease verify: [manual items from plan]\nLet me know when done.\n```\n\n**Agent Spawn:**\n```\nTask(subagent_type=\"general-purpose\", model=\"claude-opus-4-5-20251101\", prompt=\"\"\"\n[implement_task SKILL.md]\n## Context\n- Ledger: [content]\n- Plan: [section]\n- Task: [N]/[Total]: [desc]\n- Previous Handoff: [content or \"first task\"]\n- Handoff Dir: thoughts/handoffs//\n\"\"\")\n```\n\n**Recovery (post-compaction):**\n1. Ledger auto-loaded by SessionStart\n2. `ls thoughts/handoffs//`\n3. Read last handoff\n4. Resume next task\n\n---\n\n# Validity Constraints\n\n```\nforall phase. has_auto_criteria(phase) & has_manual_criteria(phase)\nforall task. one_agent_per_task(task)\nforall h. on_disk(h) -> recoverable(h)\ncompaction -> (forall h. persists(h))\nforall i < j. completed(task_i) before started(task_j)\n```","author":"@parcadei","ownerProfile":null,"authorContacts":null,"sourceUrl":"https://github.com/parcadei/Continuous-Claude-v3/tree/main/.claude/skills/implement_plan_micro","license":"MIT","category":"productivity","lang":"en","tokens":1415,"stars":0,"calls30d":1,"claimed":false,"visibility":"public","origin":"crawler","version":"0.1.0","createdAt":"2026-08-22","updatedAt":"2026-08-22","files":[],"requires":{"mcp":[],"tools":[]},"safety":{"flags":[],"scannedAt":"2026-08-22","hasScripts":false,"networkEndpoints":[]}}