bad-faith-rhetoric-detector

name	bad-faith-rhetoric-detector
description	Analyze repeated non-engagement, deflection, motte-and-bailey shifts, tone policing, sealioning, and related manipulative discourse patterns. Use when the user suspects they are being rhetorically managed rather than answered, or when they need a careful pattern read grounded in quotes and alternatives. NOT for psychiatric diagnosis, legal judgment, or help deploying bad-faith tactics.
license	Apache-2.0
allowed-tools	Read,Grep,Glob
metadata	{"category":"Communication & Analysis","tags":["rhetoric","discourse","manipulation","conflict-analysis"],"pairs-with":[{"skill":"productive-discourse-facilitator","reason":"Use it when the user wants a productive, ethical response rather than just diagnosis."},{"skill":"crisis-response-protocol","reason":"Use it first if the communication pattern suggests acute emotional or safety risk."}],"runtime":{"executionMode":"in-process","reviewSurface":"quote-table-plus-mermaid","forkWhen":["a long transcript needs one charitable read and one pattern read before naming bad faith","workplace or relationship stakes are high enough that alternative explanations must be checked independently"],"mcp-tools":[]},"provenance":{"kind":"internal-worktree-recovered","sourceRepo":"workgroup-ai","recoveredAt":"2026-04-19T06:57:25-07:00","recoveredFrom":"/Users/erichowens/coding/workgroup-ai/.claude/worktrees/agent-aa6fba4b/skills/bad-faith-rhetoric-detector","worktree":"agent-aa6fba4b","ctaOverlay":"/Users/erichowens/coding/workgroup-ai/.windags/cta-upgrades/bad-faith-rhetoric-detector/after.md","owners":["some-claude-skills"]},"authorship":{"maintainers":["some-claude-skills"]},"io-contract":{"kind":"deliverable","produces":["[Truncated]","[Truncated]","[Truncated]"]}}

Bad-Faith Rhetoric Detector

Use this skill to separate genuine disagreement from repeated rhetorical non-engagement. The core job is not to win the argument; it is to identify whether the other party is answering the substance, shifting the frame, or using the conversation itself as leverage.

When to Use

• Use this when the user says the other person keeps dodging the point, moving goalposts, or making the user defend things they already established.
• Use it when the user wants a pattern read grounded in actual quotes rather than vibes alone.
• Use it when the decision is practical: stay engaged, set a boundary, or exit.

NOT for: diagnosing personality disorders, determining legal liability, proving someone is "lying" with certainty, or helping a user weaponize manipulative tactics.

Decision Points

flowchart TD
  A[Conversation excerpt or transcript] --> B[Identify the original claim or question]
  B --> C{Does the response engage that claim directly?}
  C -->|Yes| D{Still distorted or narrowed?}
  C -->|No| E{What replaced the substance?}
  D -->|Yes| F[Motte-and-bailey, narrowing, or goalpost shift]
  D -->|No| G[Possibly disagreement, not bad faith]
  E -->|Counter-accusation| H[DARVO or deflection]
  E -->|Topic swap| I[Whataboutism or gish gallop]
  E -->|Style complaint| J[Tone policing]
  E -->|Demand for endless proof| K[Sealioning or burden shift]
  F --> L{Pattern repeated or one-off?}
  G --> L
  H --> L
  I --> L
  J --> L
  K --> L
  L -->|Repeated with clear effect| M[High-confidence pattern call]
  L -->|Ambiguous or isolated| N[Name alternatives and lower confidence]
  M --> O[Choose boundary, pause, or exit]
  N --> O

L3 Cues

• The strongest cue is not a single tactic name. It is repeated refusal to stay with the same claim long enough to evaluate it.
• Real bad-faith patterns often create asymmetry: the user must answer everything while the other person answers nothing cleanly.
• Tone policing is usually secondary. The decisive question is whether the substantive issue remained unanswered after the style complaint.
• A charitable alternative should survive contact with the quotes. If the alternative requires ignoring the transcript, it is not actually charitable.
• High-confidence calls come from recurrence plus effect: confusion, exhaustion, self-doubt, or perpetual reset of the topic.

Callsite Decomposition

1. Extract the original question or claim in the user's words.
2. Quote the response segment that allegedly failed to engage it.
3. Name the most likely tactic only after checking whether ordinary disagreement explains it.
4. Add one charitable alternative explanation.
5. State the effect on the conversation: confusion, delay, exhaustion, role reversal, or topic drift.