Skip to main content

metacognition

Checks task understanding, evidence, scope, and stopping conditions at decision-changing moments. Use when starting repository work, changing task type, encountering unexpected evidence, or completing a meaningful phase.

معلومات المصدر

المستودع
shinpr/agentic-code
آخر نشاط في المصدر
٢٩ أغسطس ٢٠٢٦ في ٠٥:٣١
لغة SKILL.md المكتشفة
الإنجليزية
النجوم
٥٠
التفرعات
٤

خيارات التثبيت

يُحدَّد Prompt الذي يراجع المصدر أولًا بشكل افتراضي. يمكنك التبديل إلى أمر مباشر أو تنزيل نسخة محلية.

مراجعة ملفات المصدر

اقرأ SKILL.md وأي ملفات مرافقة يعرضها SkillsMP قبل أن تقرر التثبيت.

عرض SKILL.md

SKILL.md
تعليمات المصدر · معاينة للقراءة فقط
name
metacognition
description
Checks task understanding, evidence, scope, and stopping conditions at decision-changing moments. Use when starting repository work, changing task type, encountering unexpected evidence, or completing a meaningful phase.
# Metacognition Protocol ## Purpose Prevent the agent from following an invalid route, expanding work without evidence, or declaring completion without observable proof. ## Checkpoints Run a checkpoint when: - repository work begins; - the task type or approved outcome changes; - unexpected evidence invalidates the current approach; - a major phase or coherent implementation task completes; - the next action requires user authority or an irreversible operation. Routine reversible actions inside an already validated approach do not require a separate checkpoint. ## Assessment ### Outcome and Scope - What observable outcome owns this work? - Which requirements are current, and which items are current-state evidence or speculation? - What non-goals and contracts must remain unchanged? - Could reuse or no-change already satisfy the outcome? ### Evidence - Which claims are observed, inferred, or unknown? - What repository rule or pattern governs the next decision? - Does an unknown change the outcome, scale, major design, authority, or verification? ### Approach - Is the current route the smallest sufficient one? - Does every added artifact, mechanism, test, gate, or task change the outcome, protect a boundary, serve a consumer, or supply necessary proof? - What current evidence would make the approach invalid? ### Completion - Is the requested result observable? - Did the applicable established checks pass? - Are verification gaps and residual limitations reported? - Did review findings remain within approved scope? ## Decision After assessment, choose one: - **continue**: Current evidence supports the route. - **adjust**: A reversible local change keeps the approved outcome intact. - **return to owning phase**: Evidence invalidates a requirement, major design decision, task boundary, or verification strategy. - **ask the user**: Progress requires a new requirement, scope change, unavailable authority, or unauthorized irreversible action. - **complete**: The outcome and required proof are observable. ## Error Recovery When an unexpected result occurs: 1. Identify the failed assumption or boundary. 2. Inspect the narrowest relevant repository evidence. 3. Prefer removing or correcting the invalid rule or implementation before adding a fallback. 4. Retry only when new evidence changes the attempt. 5. Ask the user only under the user-decision conditions above. ## Quality Check - The checkpoint changed or validated a real decision. - It did not create a status artifact solely to prove reflection occurred. - It preserved the approved outcome and stopped generating work once proof was sufficient.
عرض على GitHub