TheGoat395/Codex-Skills / skills/evidence-reconciliation
evidence-reconciliation
VerifiedResolve stale facts and authority conflicts.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 7.4 KB |
| agents/openai.yaml | config | 302 B |
| references/contradiction-protocol.md | markdown | 4.0 KB |
references/contradiction-protocol.md
View on GitHub# Contradiction Protocol
Read this before a consequential source reconciliation, large document revision, commercial decision, or research-derived operating change.
## 1. Claim ledger
Use one row for every material proposition.
| Claim | Meaning / unit | Controlling source | Other source(s) | State | Consequence | Required action |
|---|---|---|---|---|---|---|
Use the actual unit. For example, distinguish companies, contacts, messages, credits, accounts, domains, mailboxes, seats, monthly sends, and dollars. Similar nouns are not interchangeable.
Allowed states:
- **confirmed** — direct evidence agrees with the controlling source;
- **superseded** — an older source is explicitly overridden;
- **conditional** — true only when named conditions are met;
- **proposed** — a recommendation, not a current rule;
- **conflict unresolved** — sources disagree and no authority/current evidence resolves it;
- **unknown** — evidence is absent or insufficient.
### Capacity and timeline sub-ledger
For capacity/time claims, check applicable units, system boundary, limits and clock anchors. For commercial/outbound claims, use the project-governed capacity model: distinguish available credits, qualified records, delivery capacity, pacing, eligibility and actual operating time; identify the bottleneck before deriving an output or date. Do not impose sender or lead fields on unrelated systems.
## 2. Conflict types
### Authority conflict
Two sources prescribe different behavior. Follow the one with control and record the other as superseded or noncontrolling.
### Temporal conflict
The facts changed. Use the verified current state and retain the prior state only as historical context.
### Semantic conflict
The sources use similar words for different units or stages. Separate them rather than treating them as disagreement.
### Conditional conflict
Both sources can be true under different conditions. State the conditions explicitly and do not turn either into a universal rule.
### Path-selection conflict
Similar facts can belong to mutually exclusive paths. Select the actual path before combining timelines, costs, capacities, or prerequisites. Do not use an attribute of one path to prove a conclusion about another.
### Evidence-quality conflict
Marketing, commentary, or estimates disagree with official or direct evidence. Keep the weaker source as hypothesis/context, not as an operating fact.
## 3. Source-to-output matrix
Before a multi-document rewrite, create this map:
| Output document / section | Existing source sections preserved | New authoritative updates | Claims removed or relabeled | Verification |
|---|---|---|---|---|
Every standing document must have a row. If a source is not carried forward, record why.
## 4. Non-regression test
Before delivery, compare the revision to the standing source:
- Does it preserve the original document's practical job?
- Did any unique procedure, template, scenario, exception, price condition, ownership rule, or recovery step disappear?
- Did a short summary replace usable operating detail?
- Did a new tool, feature, or service become a promise without an explicit gate?
- Did a contradictory number remain unqualified?
- Did a trial allowance get treated as usable after an upgrade or after its expiry without authenticated proof?
- Did a capacity or deadline statement name a clock anchor and respect every relevant lead, credit, delivery, health, and operating gate?
If yes, repair it or disclose it as a deliberate removal with the controlling reason.
## 5. Decision wording
Use one of these forms:
- **Current rule:** “The controlling source currently requires …”
- **Verified fact:** “The authenticated/current official surface shows …”
- **Conditional:** “This applies only if …”
- **Recommendation:** “I recommend … because …; it is not yet adopted.”
- **Unresolved:** “The sources differ. The controlling source is …; this needs … before it can change.”
Never use a confident conclusion where the correct state is conditional or unresolved.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |