AI agent skill · by @zhnnky329

Modeler Decision Logger

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.

Archived 2026-10-23 · Source updated 2026-09-24 · License: MIT. Check the original repository for the latest version.

SKILL.md instructions

# 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

```json
{
  "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.