How do I install the Metacognitive Plan skill?

Run npx skillmds@latest add lstr-1/metacognitive-plan in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Metacognitive Plan safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Metacognitive Plan?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Metacognitive Plan free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Metacognitive Plan?

LSTR-1 (@lstr-1) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Metacognitive Plan by lstr-1 — Agent Skill

Name: Metacognitive Plan
Author: LSTR-1

Metacognitive Plan

Orchestrates the LSTR 6-step metacognitive reasoning chain and produces a decision-complete plan against verifiable constraints. Invoke for architecture decisions, multi-constraint trade-offs, incident triage, or any task requiring auditable reasoning. Trigger phrases "plan this", "/plan" (delegating to the reasoning chain), "apply the LSTR framework", "reason through this", or any query where constraints compete and a plan is needed.

LSTR-1 47bf83c 2.5 KB Updated Sep 17, 2026

File contents

metacognitive-plan

Runs the LSTR 6-step chain (delegating the exact tool-argument contract to the global lstr-reasoning-framework skill), then adds constraint extraction and mental simulation before committing to a plan.

When to Use

Multi-step inference, competing constraints, non-trivial uncertainty, or material consequence (≥2 complexity-gate criteria).
Any query where an auditable reasoning trail is required.

Instructions

Complexity gate — evaluate against 10-reasoning-framework.md. If <2 criteria, skip the chain and state "framework not warranted".
Run the 6-step chain via the lstr-reasoning MCP family, per lstr-reasoning-framework:
- metacognitiveMonitoring (knowledge-assessment) → sequentialthinking (≥3 thoughts) → collaborativeReasoning (≥2 divergent personas) → scientificMethod + structuredArgumentation (thesis + objection + synthesis) → constraintSolver (numeric only).
- EXCLUDE narrative-planner.
Constraint extraction (DeepPlanning) — classify every constraint as local (step-scoped) or global (cross-subtask budget / dependency). Record each constraint's verifiability: deterministic, rule-checkable, or qualitative.
Mental simulation (Qwen-AgentWorld) — before committing, rehearse the top candidate plan against the environment model; predict where it fails under resource limits or silent degradation; backtrack if a global constraint is violated.
Emit a decision-complete plan — ordered, dependency-aware steps, each with an exit condition. No calendar dates; use predecessor relationships.

Output Contract

Decision-complete plan (executor makes no additional choices).
Local vs global constraint list, each with verifiability label.
Reasoning record persisted per lstr-reasoning-framework contract.
Confidence from metacognitiveMonitoring.overallConfidence.

Constraints

Do NOT fabricate a reasoning trace. A failed step is logged failed.
Do NOT encode qualitative constraints as constraintSolver booleans.

---name: metacognitive-plandescription: Orchestrates the LSTR 6-step metacognitive reasoning chain and produces a decision-complete plan against verifiable constraints. Invoke for architecture decisions, multi-constraint trade-offs, incident triage, or any task requiring auditable reasoning. Trigger phrases "plan this", "/plan" (delegating to the reasoning chain), "apply the LSTR framework", "reason through this", or any query where constraints compete and a plan is needed.---# metacognitive-planRuns the LSTR 6-step chain (delegating the exact tool-argument contract to the global `lstr-reasoning-framework` skill), then adds constraint extraction and mental simulation before committing to a plan.## When to Use- Multi-step inference, competing constraints, non-trivial uncertainty, or material consequence (≥2 complexity-gate criteria).- Any query where an auditable reasoning trail is required.## Instructions1. **Complexity gate** — evaluate against `10-reasoning-framework.md`. If <2 criteria, skip the chain and state "framework not warranted".2. **Run the 6-step chain** via the lstr-reasoning MCP family, per `lstr-reasoning-framework`: - `metacognitiveMonitoring` (knowledge-assessment) → `sequentialthinking` (≥3 thoughts) → `collaborativeReasoning` (≥2 divergent personas) → `scientificMethod` + `structuredArgumentation` (thesis + objection + synthesis) → `constraintSolver` (numeric only). - EXCLUDE `narrative-planner`.3. **Constraint extraction (DeepPlanning)** — classify every constraint as `local` (step-scoped) or `global` (cross-subtask budget / dependency). Record each constraint's verifiability: deterministic, rule-checkable, or qualitative.4. **Mental simulation (Qwen-AgentWorld)** — before committing, rehearse the top candidate plan against the environment model; predict where it fails under resource limits or silent degradation; backtrack if a global constraint is violated.5. **Emit a decision-complete plan** — ordered, dependency-aware steps, each with an exit condition. No calendar dates; use predecessor relationships.## Output Contract- Decision-complete plan (executor makes no additional choices).- Local vs global constraint list, each with verifiability label.- Reasoning record persisted per `lstr-reasoning-framework` contract.- Confidence from `metacognitiveMonitoring.overallConfidence`.## Constraints- Do NOT fabricate a reasoning trace. A failed step is logged `failed`.- Do NOT encode qualitative constraints as `constraintSolver` booleans.

Frequently asked questions