← Back to Skills Marketplace
lawliet-ai

First-Principles-Deconstructor

by Lawliet-ai · GitHub ↗ · v1.0.1 · MIT-0
cross-platform ⚠ pending
320
Downloads
0
Stars
0
Active Installs
2
Versions
Install in OpenClaw
/install first-principles
Description
Analyzes products by stripping marketing jargon to identify their fundamental physical laws, constraints, and optimal efficient designs from first principles.
README (SKILL.md)

Skill: First-Principles-Audit (Truth-Filter V2.1)

1. Core Assertion

System SHALL systematically strip all "Marketing Jargon" and "Industry Inertia." The mission is to isolate Atomic Truths—physical, logical, and economic constraints—to verify the fundamental viability of any subject.

2. Operational Phases

Phase 0: Technical Ground-Truth Retrieval

  • Mandate: Locate patents, technical white papers, cost-structure breakdowns, and raw physical constants.
  • Filtering: Explicitly ignore PR announcements, celebrity endorsements, and "Industry Standard" claims.

Phase 1: Inescapable Constraint Mapping

  • Assertion: Identify 3-5 Atomic Constraints (e.g., Energy Density, Marginal Cost, Bandwidth Limits, Entropy).
  • Classification: Categorize every feature as either an "Immutable Constant" or "Marketing Fluff."

Phase 2: Efficiency-Logic Comparison

  • Action: Perform a "From-Scratch" reconstruction.
  • Logic: Do not compare feature lists. Compare the Efficiency of Implementation Logic.
  • Metric: If a solution deviates from the most efficient physical path, it MUST be flagged as a logical anomaly.

Phase 3: Logical Anomaly Detection (The Verdict)

  • Action: Generate a "Hard-Audit Verdict."
  • Detection: Explicitly point out where the current solution violates physics, economics, or basic common sense.

Phase 4: Chaining & Cross-Verification (Critical)

  • Action: Consume data from ~/.openclaw/swarm_tmp/expert_output.json if available.
  • Serialization: Map findings into the local audit_schema.json.
  • Persistence: Write the final report to ~/.openclaw/swarm_tmp/audit_report.json.

3. Hard Constraints

  • C1 (Zero-Tolerance): Use of marketing terms (e.g., "Revolutionary," "Leading-edge") SHALL trigger a mandatory rewrite.
  • C2 (Evidence-First): Every audit claim MUST be backed by a physical or logical constant.
  • C3 (Structured Output): No conversational filler. Output MUST be a valid JSON payload following the Phase 4 mandate.
Usage Guidance
This skill appears coherent and low-risk: it does not install code or request credentials. Before installing, consider: 1) The skill will read ~/.openclaw/swarm_tmp/expert_output.json if that file exists — review that file for any sensitive content you would not want the skill to consume. 2) It will write a machine-readable audit to ~/.openclaw/swarm_tmp/audit_report.json — ensure that location is acceptable and not shared. 3) The agent will likely fetch public patents, papers, and web resources during use; if you restrict network access you may limit functionality. 4) Because outputs are required to be strict JSON and evidence-backed, verify the agent includes citations or references when you need source traceability. If the package later adds code files, external download/install steps, or requests credentials, re-evaluate (those changes could raise suspicion).
Capability Assessment
Purpose & Capability
Name/description (strip marketing, find physical constraints, produce a verdict) matches the instructions. Required actions (search patents/whitepapers, compute constraints, produce JSON audit) are coherent with an audit/deconstruction tool.
Instruction Scope
Instructions are focused on public-knowledge retrieval and logical/physical analysis. They do, however, explicitly tell the agent to read ~/.openclaw/swarm_tmp/expert_output.json if present and to write results to ~/.openclaw/swarm_tmp/audit_report.json and to map output to the included audit_schema.json. Reading and writing that local path is consistent with chaining/automation but means any pre-populated expert_output.json will be consumed; the skill otherwise does not instruct reading other unrelated system files.
Install Mechanism
No install spec and no code files beyond static schema and docs. Instruction-only skills have the lowest install risk; nothing will be downloaded or executed as part of an install step.
Credentials
The skill declares no environment variables, no credentials, and no config paths beyond its own ~/.openclaw/swarm_tmp files and the included audit_schema.json. This is proportionate to its audit purpose.
Persistence & Privilege
always is false (no forced inclusion). The only persistence is writing a report into ~/.openclaw/swarm_tmp/audit_report.json and reading a possible expert_output.json in the same directory. This is reasonable for chaining but does require write/read access to that local path.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install first-principles
  3. After installation, invoke the skill by name or use /first-principles
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.1
- Introduced formal phase-driven methodology with stricter language and explicit mandates. - Added a new Phase 4: Chaining & Cross-Verification, requiring use of external expert output and serialization of audits into `audit_schema.json`. - Defined hard constraints enforcing zero marketing language, evidence-backed claims, and structured JSON output only. - Updated skill name from "First-Principles-Deconstructor" to "First-Principles-Audit (Truth-Filter V2.1)". - Added new file: `audit_schema.json` for report structure compliance.
v1.0.0
Specifically designed for hardware research, technical comparisons, and business feasibility analysis. This skill drives AI to penetrate surface-level parameter comparisons, automatically searching for underlying physical limitations, cost structures, and core algorithms. It serves as a "truth scanner" for product selection experts, architects, and decision-makers, aimed at identifying what constitutes true innovation versus premium jargon.
Metadata
Slug first-principles
Version 1.0.1
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 2
Frequently Asked Questions

What is First-Principles-Deconstructor?

Analyzes products by stripping marketing jargon to identify their fundamental physical laws, constraints, and optimal efficient designs from first principles. It is an AI Agent Skill for Claude Code / OpenClaw, with 320 downloads so far.

How do I install First-Principles-Deconstructor?

Run "/install first-principles" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is First-Principles-Deconstructor free?

Yes, First-Principles-Deconstructor is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does First-Principles-Deconstructor support?

First-Principles-Deconstructor is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created First-Principles-Deconstructor?

It is built and maintained by Lawliet-ai (@lawliet-ai); the current version is v1.0.1.

💬 Comments