CBT CardsFind a practice
CBT Cards / Toolkit / Audit

115 records are not 115 equally useful ideas.

The pinned source corpus is now triaged record-by-record before any further CBT Cards publication. The audit makes duplication, framework mixing, treatment-like wording, and editorial priorities explicit.

Record-complete audit

A work queue, not a popularity contest.

6 published

Already covered by the separate CBT Cards publication-review overlay and canonical standalone pages.

31 card candidates

Source cards distinct enough to merit editorial review. Candidate status is not approval.

12 metaphor candidates

Potential memory aids that still need pairing, scope, and wording review. Metaphors are not evidence.

15 need rework

Cards with treatment-like, suppression, attribution, avoidance, overclaim, or mechanism problems that should not be promoted as-is.

15 protocol-gated

Every protocol remains behind a separate protocol-level evidence and safety review.

36 source-only holds

Mostly overlap, generic wellness material, or lower incremental value relative to the reviewed owned-practice layer.

What the audit found

Sixty-five source records carry at least one quality flag and seven explicit similarity groups capture repeated ideas. Exact duplicate titles include Shift Your Focus (cards 24 and 54), See It Differently (cards 36 and 68), and Burnout Recovery Protocol (protocols 1 and 8). Stable source IDs therefore remain authoritative.

The corpus also mixes CBT, ACT-adjacent, DBT-adjacent, generic wellness, metaphor, and mixed-protocol material. Source inclusion is not enough to relabel all of it “CBT.”

Records that should not be promoted as-is

The audit blocks a small set of source cards from the normal candidate queue because the wording or mechanism needs substantial rework. Examples include generic affirmations, “gratitude rewires your brain” wording, distraction framed as the default answer to repetitive thinking, “fake it till you make it,” unsupervised exposure-like prompts, opposite-action wording without a danger check, ACT/DBT techniques presented without framework attribution, thought stopping, ERP, and emotion-exposure wording.

That does not mean every underlying mechanism is useless. It means the raw card is not a safe publication-ready contract.

Protocols get their own gate

All 15 protocol records remain source_only for CBT Cards publication. Multi-step protocols can imply a treatment sequence more easily than a small reflection prompt, so they require a separate methodology, evidence, safety, and claims review rather than being promoted through the ordinary card pipeline.

First review queue

The first distinct candidates worth deeper review are deliberately modest: card-2 Spot the Distortion, card-25 What Set You Off?, card-31 Thought or Feeling?, card-35 Name That Emotion, card-43 What Matters Most?, card-45 Is It Worth It?, card-69 Understand the ABCs, and card-70 Break Down the Behavior. Four metaphors are queued only as possible memory aids: metaphor-4, metaphor-5, metaphor-18, and metaphor-19.

Machine-readable audit

The complete triage partition, framework buckets, quality flags, similarity groups, owned-practice overlaps, and review queue are published in data/toolkit-audit.json. The audit covers every pinned source ID exactly once.

Publication boundary

This audit is not publication approval. Only data/toolkit-review.json can mark a raw source record reviewed_for_publication and published. Everything else remains source-only until an explicit review decision creates a canonical CBT Cards resource.