How do I install the Consistency Auditor skill?

Run npx skillmds@latest add zhnnky329/consistency-auditor 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 Consistency Auditor 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 Consistency Auditor?

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 Consistency Auditor free to use?

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

Who published Consistency Auditor?

zhnnky329 (@zhnnky329) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Consistency Auditor by zhnnky329 — Agent Skill

Name: Consistency Auditor
Author: zhnnky329

Consistency Auditor

Run scoped or final cross-media consistency checks for mathematical-modeling artifacts, comparing canonical numbers, symbols, parameters, decisions, files, and paper claims without performing full-workspace audits for low-risk changes.

zhnnky329 Updated Sep 17, 2026

File contents

Modes

scoped: run only for a CANONICAL or FROZEN change and only for affected Qx/identifiers.
final: run across all submission artifacts before assembly.

Sources of Truth

Prefer:

frozen_numbers.json for paper numbers;
qx_decisions.jsonl for human judgments;
planning/symbol_table.md for symbols/units;
approved code plan and run summary for parameters and executed methods;
solution package for writer-facing structure.

Use legacy decision logs or Markdown reviews only during migration.

Checks

Numerical claims match frozen values and units.
Formula, parameter, constraint, and method roles match approved code and plan.
Symbols and units match the global symbol table.
Referenced tables, figures, code, and data files exist.
Why-this-method, result verdict, stability, confidence, and claim scope resolve to human decision IDs.
Freeze and decisions are not stale relative to materially changed cited evidence.

Scoped Workflow

Receive changed files/identifiers and their impact class.
Resolve affected Qx and downstream consumers.
Check only those relationships.
Return a compact digest or save results/Qx/reports/scoped_consistency.json when durable evidence is needed.
Do not rewrite other artifacts.

Final Workflow

Check all submission Qx.
Save paper/audits/cross_media_consistency_audit.md.
Report PASSED, FAILED, or NOT_RUN.
List concrete divergences and repair owners. Do not pad a pass list.

Rules

Do not run a full audit for formatting, comments, scratch work, or ordinary pre-freeze exploration.
Do not infer canonical numbers from the paper.
Do not repair divergences inside the audit.
Do not approve final assembly directly.

Verification

Scope matches the semantic impact.
Every divergence identifies source, consumer, and expected repair.
Final audit covers all six check classes.
Verdict follows actual evidence rather than an item count.

---name: consistency-auditordescription: Run scoped or final cross-media consistency checks for mathematical-modeling artifacts, comparing canonical numbers, symbols, parameters, decisions, files, and paper claims without performing full-workspace audits for low-risk changes.---# Modes- `scoped`: run only for a `CANONICAL` or `FROZEN` change and only for affected Qx/identifiers.- `final`: run across all submission artifacts before assembly.# Sources of TruthPrefer:- `frozen_numbers.json` for paper numbers;- `qx_decisions.jsonl` for human judgments;- `planning/symbol_table.md` for symbols/units;- approved code plan and run summary for parameters and executed methods;- solution package for writer-facing structure.Use legacy decision logs or Markdown reviews only during migration.# Checks1. Numerical claims match frozen values and units.2. Formula, parameter, constraint, and method roles match approved code and plan.3. Symbols and units match the global symbol table.4. Referenced tables, figures, code, and data files exist.5. Why-this-method, result verdict, stability, confidence, and claim scope resolve to human decision IDs.6. Freeze and decisions are not stale relative to materially changed cited evidence.# Scoped Workflow1. Receive changed files/identifiers and their impact class.2. Resolve affected Qx and downstream consumers.3. Check only those relationships.4. Return a compact digest or save `results/Qx/reports/scoped_consistency.json` when durable evidence is needed.5. Do not rewrite other artifacts.# Final Workflow1. Check all submission Qx.2. Save `paper/audits/cross_media_consistency_audit.md`.3. Report `PASSED`, `FAILED`, or `NOT_RUN`.4. List concrete divergences and repair owners. Do not pad a pass list.# Rules- Do not run a full audit for formatting, comments, scratch work, or ordinary pre-freeze exploration.- Do not infer canonical numbers from the paper.- Do not repair divergences inside the audit.- Do not approve final assembly directly.# Verification- Scope matches the semantic impact.- Every divergence identifies source, consumer, and expected repair.- Final audit covers all six check classes.- Verdict follows actual evidence rather than an item count.

Frequently asked questions