Skip to content
alyn
zhnnky329

Modeler Decision Logger skill

Faithfully append a human modeler's choice and rationale to one canonical per-subquestion JSONL decision ledger. Use after a choice card is answered or when migrating legacy decision artifacts; never originate or improve the decision.

$npx skills add zhnnky329/MathModeling-skills --skill modeler-decision-logger
View this skill on GitHubzhnnky329/MathModeling-skills/.claude/skills/modeler-decision-logger/SKILL.md
Pack stars
1,088
Pack installs
1k
Alyn Score
49/99
Updated
2 months ago
License
MIT
– views – installs via Alyn
AI engineering

Install Modeler Decision Logger

Pick your agent. Every command is ready to paste.

$npx skills add zhnnky329/MathModeling-skills --skill modeler-decision-logger

The open skills CLI detects your agents (Claude Code, Cursor, Codex and 70+ more) and installs to the right folder. Add -g to install globally.

01

Install it

Run the command above, or upload the folder in Claude.

02

Ask naturally

Describe the task. Your agent loads modeler-decision-logger when the request matches its description.

03

Name it if needed

If it does not trigger, say “use the modeler-decision-logger skill”. In Codex, type $modeler-decision-logger.

The skill itself

SKILL.md

FolderGitHub

Purpose

Make human judgment traceable without multiplying decision files.

Canonical Output

methods/Qx/qx_decisions.jsonl

One JSON object per line. Records are append-only.

Use planning/framing_decisions.jsonl for global/pre-Qx framing decisions.

Required Fields

{
  "schema_version": 1,
  "decision_id": "q1_method_choice",
  "decision_type": "method_choice",
  "status": "DECIDED",
  "decided_by": "human",
  "captured_in_mode": "learning",
  "choice": "M2",
  "rationale": "Human-authored reason tied to evidence.",
  "evidence_refs": ["methods/Q1/probes/risk_probe_summary.json"],
  "decided_at": "ISO-8601",
  "supersedes": null
}

Optional structured fields may include confidence, rejected alternatives, round action, claim scope, assumption labels, or fallback activation.

Workflow

  1. Receive the human's answer, the choice-card ID, and evidence paths.
  2. Preserve the user's meaning and wording. Normalize only structure, identifiers, and whitespace.
  3. Verify:
    • the choice is one of the presented options or explicitly records a user-supplied alternative;
    • evidence paths exist;
    • rationale is non-empty and contains no placeholder;
    • the record does not falsely label AI-authored prose as human-authored.
  4. Append one JSON line.
  5. If revising a decision, append a new record with supersedes; never overwrite history.
  6. Update the compact history in qx_method_card.md only when the decision changes method state.
  7. Update the manifest gate/status fields when present.

Decision Types

Typical values:

  • framing
  • method_choice
  • fallback_activation
  • result_verdict
  • stability_verdict
  • assumption_necessity
  • claim_scope
  • package_signoff
  • submission_authorization

Staleness

Mark a decision stale only when its cited evidence materially changed:

  • append a decision_stale record naming the old decision and changed evidence;
  • ask the human to reconfirm through one choice card;
  • do not mark decisions stale because unrelated files or formatting changed.

Legacy Migration

Read legacy:

  • methods/Qx/qx_decision_log.md
  • methods/Qx/decisions/*_modeler_decision.md

Migrate only completed human decisions. Preserve original timestamps and source paths when available. Do not convert PENDING placeholders into decisions.

Rules

  • Never choose, rationalize, strengthen, or complete the user's decision.
  • Never create a separate pending decision artifact.
  • Never copy ai_suggestion into the human rationale.
  • One evidence-linked sentence is enough; do not impose arbitrary prose length.
  • Preserve an honest AI-use provenance distinction.

Verification

  • JSONL is valid one-object-per-line.
  • Record is append-only and uniquely identified.
  • Human ownership and evidence are accurate.
  • Supersession and staleness preserve history.

Tips for Modeler Decision Logger

  • • Start a fresh session after installing so your agent picks up the new skill.
  • • It is part of a pack with 29 skills. Install only the ones you need with --skill to keep triggers precise.
  • • If two skills overlap, mention the one you want by name.

Questions about Modeler Decision Logger

What does the Modeler Decision Logger skill do?

Faithfully append a human modeler's choice and rationale to one canonical per-subquestion JSONL decision ledger. Use after a choice card is answered or when migrating legacy decision artifacts; never originate or improve the decision.

How do I install Modeler Decision Logger?

Run "npx skills add zhnnky329/MathModeling-skills --skill modeler-decision-logger" in your project. The skills CLI installs it for Claude Code, Cursor, Codex and 70+ other agents. In Claude on the web, upload the skill folder as a ZIP under Customize > Skills.

Does Modeler Decision Logger work in Cursor and Codex?

Yes. Modeler Decision Logger uses the open SKILL.md format, so it works in Claude Code, Claude, Cursor, OpenAI Codex, Gemini CLI, GitHub Copilot and other agents that support agent skills.

Is Modeler Decision Logger free?

Modeler Decision Logger is open source on GitHub under the MIT license. It is free to install; check the license before commercial use.

Who made Modeler Decision Logger?

Modeler Decision Logger is published by zhnnky329 in the zhnnky329/MathModeling-skills repository, together with 28 other skills.

Similar skills. From other creators.