Home / All Skills / Workflow & Core Tools / ljg-constraint
Workflow & Core Tools · lijigang/ljg-skills

ljg-constraint

Identify the defining constraints that genuinely bound a domain, profession, role, product, or debate; categorize them into world/rule/interpretation (hard/soft/self-imposed); and examine how this set of constraints defines identity, completes the problem statement, delineates the solution space, and explains observed real-world behaviors. Especially useful for distinguishing strategic disagreements where goals are identical but default constraints differ, and for identifying outdated interpretations mistaken for hard facts. USE WHEN the user says 'constraint', 'find constraints', 'what are the constraints of this domain', 'essential constraint', 'what bounds it', 'hard constraint or self-imposed constraint', 'where is the boundary', 'why does this industry do this', 'constraint', '/ljg-constraint', or provides any domain/profession/role/product/debate wanting its defining constraints and solution space mapped. NOT FOR drilling down into generative mechanisms (use ljg-rank to find generative power rather than boundaries), drilling deep vertically into an argument (use ljg-think), breaking down a book (use ljg-book), or dissecting a concept (use ljg-learn).

L2 Local · Restricted workspaceV1 Commit-PinnedWrite FileDual-pass Audit
Author lijigangGitHub lijigang/ljg-skills ↗License MIT(仓库根 LICENSE)Commit 9e75497a95
Control levels describe skill capabilities and suggested control measures; they do not represent quality or safety certifications, nor do they dictate usability. Usage decisions rest entirely with users and administrators. Different agent hosts apply varying constraints; do not assume hosts provide automatic protection. Third-party skill copyrights belong to their original authors. Use at your own risk.

1How It Works & Mechanism

Constraints are not negative shackles; they are the system of equations that collapses infinite possibilities into 'this specific one'. The goal and the constraints together constitute the full problem statement.

skills/ljg-constraint/SKILL.md
约束不是在减少可能性,是从无穷里生出特定性。
Purpose here: Reset the colloquial intuition of 'constraint', repositioning the search for constraints as describing identity rather than listing difficulties.

Three levels of hardness: hard constraints (world/physical level, violating causes system collapse), soft constraints (rule level, violating incurs costs), and self-imposed constraints (interpretation/cognitive level, can be redefined).

skills/ljg-constraint/SKILL.md
**硬约束(世界层 / 物理层)**——不可违背,试图违背则系统崩溃。
Purpose here: Assign hardness to each candidate constraint as the first step in accurately characterizing its true weight.

Execute the seven-step process (list candidates / layer and position / determine authenticity / identify constraint mismatches / identify contradictory pairs / identify identity constraints / synthesize via two lenses) mentally only; do not write it into the final article.

skills/ljg-constraint/SKILL.md
工序心里走完,别写进文章——读者看不到你怎么摸的边,但感觉得到底下是摸到实处的。
Purpose here: Separate the internal process from the final artifact—readers only read the chain of conclusions, not the exploratory probing.

Anchor conclusion in explanatory power: Rational, optimal behavior within the solution space should align precisely with what the role repeatedly does in reality for the constraints to be considered accurately identified.

skills/ljg-constraint/SKILL.md
**解释实际操作。** 这是收口,也是这引擎最值钱的地方:解空间里那个理性的最优行为,应该正好就是这个角色在现实里反复做的那些事。
Purpose here: Translating 'constraints → solution space' into explanations of concrete behaviors is the most valuable deliverable of this engine.

The first gate and primary criterion of the post-check: Explanatory power—whether the rational, optimal behavior in the solution space matches observed real-world practice.

skills/ljg-constraint/SKILL.md
1. **解释力(主判据)**——解空间里的理性最优行为,对不对得上现实中观察到的实际操作?
Purpose here: If explanatory power fails to align, return to the process to dig deeper, indicating a missed constraint or a misclassified hardness level.

'How to improve' is secondary and does not participate in diagnosis: append only one or two sentences on potential leverage points after the diagnosis is complete, without retroactively altering the diagnostic findings.

skills/ljg-constraint/SKILL.md
**急着找出口,就会看不清墙**——这是这引擎最容易犯的病,务必压住。
Purpose here: Suppress the urge to reform until after description is complete, preventing premature optimism from biasing leverage points.

Native Chinese anti-collapse gate: silently ask before writing, 'Would a native speaker who has never read English phrase it this way?', and rewrite the entire paragraph if translationese or anglicisms are detected.

skills/ljg-constraint/SKILL.md
**「一个没读过英文的中国人,会这么说话吗?」**
Purpose here: Use a single self-check question to keep translationese and nominalized abstraction out of the text.

Append an ASCII structural diagram at the end, shaped by the analytical framework; strictly prohibit all Unicode symbols (arrows, boxes, bullets, etc.) in the hard constraint layer.

skills/ljg-constraint/SKILL.md
**硬约束:只用纯 ASCII 字符。** 禁用任何 Unicode 符号(箭头 → ← ↑ ↓、方框 ┌─┐└┘│、圆点 • ● 等)。
Purpose here: Clarify the shape of the relationship between constraints and behavior before selecting a drawing style among three available archetypes.

Output workflow: Obtain timestamp, write to ~/Documents/notes/{timestamp}--{domain}的约束__constraint.org with frontmatter following references/template.org, and report the path.

skills/ljg-constraint/SKILL.md
写入 `~/Documents/notes/{时间戳}--{领域}的约束__constraint.org`,frontmatter 见 `references/template.org`
Purpose here: Establish the artifact format (org-mode note) and local storage destination.

Gotchas baseline: Never write constraints as ordinary difficulties, flaws, or advice; a constraint must alter the solution space such that removing it changes permissible behaviors.

skills/ljg-constraint/SKILL.md
不要把约束写成普通困难、缺点或建议。约束必须能改变解空间;拿掉它,允许的行为会变。
Purpose here: Provide counter-criteria for 'what does not count as a constraint' to prevent falling into generic difficulty lists.

2Core Capabilities

01Identify essential constraints for a domain/profession/role/product and classify them by hardness (hard/soft/self-imposed)
02Determine authenticity: Identify outdated interpretations mistaken for hard facts, demoting them via the historical gate (whether other contexts or eras have bypassed them)
03Analyze constraint mismatches: Explain strategic disagreements where parties share identical goals but operate under different default constraints, creating different problem statements
04Delineate the solution space from constraints and explain actual behavior (grounded in explanatory power as the primary criterion)
05Analyze contradictory pairs: Explain awkward or seemingly irrational behaviors when mutual exclusion among constraints collapses the solution space
06Identify identity constraints: Find the defining boundary without which the role ceases to be itself
07Synthesize via two lenses: Function theory (constraint = domain of definition) and evolutionary theory (constraint = selection pressure)
08Append terminal ASCII diagrams: Explanatory chain diagram / Hardness nesting diagram / Constraint contradiction diagram
09Clear division of labor with ljg-rank: ljg-rank digs downward for generative power; constraint reaches outward for boundaries

4Recommended Governance & Evidence

Suggested Controls for Safe Operation
For Agent Runtime
Restrict working directory
For Administrators
Retain operation logs
Evidence & Triggers (File Paths & Excerpts)
  • 本地工作目录: File I/O and artifact output: Writes to `~/Documents/notes/{timestamp}--{domain}的约束__constraint.org` per SKILL.md output section (handled by host/agent).
Audit Notes: Evaluation target: this skill directory (skills/ljg-constraint/: SKILL.md + references/template.org). Verification: zero scripts, zero network endpoints, zero credential access within the directory; risk scan hit 0 across all 7 categories. Classified as Blue instead of Green solely because the SKILL.md output section directs writing analysis to `~/Documents/notes/` as an org file, constituting local write operations (Rule: Green = no local writes, Blue = local read/write only). The write operation targets plain-text notes under a fixed path in the user's Documents directory, with no external egress or credential handling, requiring no tier elevation.

5Second-Pass Verification

  • [ok] 文件构成 — §1 清单仅 SKILL.md 与 references/template.org,无脚本、无资产、无端点。
  • [ok] 三层硬度与工序连贯 — 硬/软/自设三层定义与七步工序、心内判据互相一致,解释力被反复点为主判据。
  • [ok] 描述优先于改造 — 「松动哪条(第二位,不参与诊断)」与 Gotchas「不要急着写如何突破约束」一致。
  • [ok] ASCII 图约束 — 文末结构图规定硬约束层禁用 Unicode 符号,并给出三种可选形态。
  • [ok] 风险扫描复核 — §3 七类(URL/凭证/执行/写盘/网络/危险旗标/沙箱)均 0 命中。
  • [ok] 判级口径 — 纯指令 + 一处本地写入(org 笔记)→ 按范式落 blue,非 green。

6Verdict & Scenarios

  • Codifies constraint discovery into a reusable epistemological workflow: hardness stratification, authenticity assessment, constraint mismatch, contradictory pairs, identity constraints, and dual lenses, with explanatory power as the falsifiable primary benchmark.
  • Strictly decouples diagnosis from intervention (leverage points are secondary and excluded from diagnosis), preventing analysis from being derailed by problem-solving impulses.
  • Includes ASCII structural diagrams and native language anti-collapse quality gates, balancing conceptual visualization with high prose quality.
  • Best for: Users who need to clearly see what bounds a domain, profession, role, or product; want to explain seemingly odd or irrational behaviors; or need to deconstruct why a solution debate remains deadlocked. Also suitable for identifying outdated interpretations mistaken for hard facts.
    Not for: Digging into generative mechanisms (use ljg-rank), drilling deep into a single argument (use ljg-think), breaking down a book (use ljg-book), or dissecting a concept (use ljg-learn); also unsuitable for requests merely seeking a tactical 'how-to' list for overcoming a constraint.
    Install Agent-Ready
    ① Verified Mirror 2026-10-10
    Option A · Direct DownloadDownload ljg-constraint.tar.gz
    sha256: f61adce608c1df97…
    Option B · Copy Guide for Agent
    install.json
    Agents automatically download and verify mirrors from this JSON guide.
    ② Upstream GitHub
    GitHub Repo ↗
    Pinned to commit 9e75497a95.