TheGoat395/Codex-Skills / skills/logical-thinking
logical-thinking
VerifiedDetermine what premises and evidence warrant.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 4.9 KB |
| agents/openai.yaml | config | 318 B |
| references/causal-probabilistic-decision.md | markdown | 4.9 KB |
| references/documentation-evidence-conclusions.md | markdown | 3.9 KB |
| references/formal-defeasible-inference.md | markdown | 5.3 KB |
| references/proof-and-verification.md | markdown | 4.5 KB |
references/documentation-evidence-conclusions.md
View on GitHub# Documentation And Evidence Conclusions
Read this reference when drawing conclusions from policies, contracts, specifications, manuals, source code, tests, benchmarks, account state, logs, or runtime observations.
## Classify the source statement
Determine whether the source provides:
- definition;
- mandatory requirement;
- prohibition;
- permission or capability;
- default behavior;
- recommendation;
- example;
- limit or threshold;
- marketing claim;
- observed result;
- proposed future behavior.
Do not turn `may` into `will`, a default into a requirement, an example into exhaustive behavior, or a recommendation into a governing rule.
## Establish applicability
Record:
- authority and whether the source controls the question;
- publication, effective, version, commit, or observation date;
- product, plan, jurisdiction, environment, branch, or account scope;
- prerequisites and feature flags;
- exceptions, precedence, and supersession;
- whether the text describes intended behavior or observed behavior.
If authority or sources conflict, use `evidence-reconciliation` before concluding.
## Match conclusion to verification layer
| Evidence | Warranted conclusion |
|---|---|
| Governing text | The rule requires, permits, or prohibits X within its scope |
| Product documentation | X is documented for the stated version and conditions |
| Source code | The inspected code implements path X, subject to build/runtime configuration |
| Configuration | X is structurally configured, not necessarily active |
| Passing build/static check | The checked artifact satisfies that check |
| Targeted runtime test | X occurred in the tested environment and case |
| Production observation | X was observed live at the recorded time and scope |
| Benchmark | Candidate A performed as measured under the benchmark protocol |
Do not infer live operation from documentation, universal behavior from one test, production readiness from a build, or general superiority from one benchmark.
## Resolve a documentary conclusion
Use:
```text
Question:
Controlling source and version:
Exact applicable rule or behavior:
Prerequisites/exceptions:
Observed implementation state:
Conflict or gap:
Strongest warranted conclusion:
Recheck trigger:
```
Quote only the minimum exact language needed. Preserve definitions and operative qualifiers.
## Verify independently
Ask verification questions without embedding the draft answer:
- Which source controls this situation?
- What version and scope apply?
- What must be true before the rule activates?
- What exceptions or precedence clauses exist?
- What observable evidence would show implementation?
- Does current state match the rule?
For a benchmark, independently check prompt, model/version, parameters, dependencies, hardware, repetitions, scoring, exclusions, uncertainty, and raw artifacts.
## Conclusion forms
- **Governing:** an applicable authority settles the rule.
- **Documented:** official text describes the capability or behavior.
- **Implemented:** source/configuration structurally contains it.
- **Observed:** direct evidence shows it occurred in a bounded case.
- **Verified in scope:** a suitable test establishes the stated behavior under recorded conditions.
- **Unsupported:** the source does not establish the proposed claim.
- **Applicability unresolved:** controlling scope, version, or prerequisite remains uncertain.
Keep these forms separate in the final wording.
## Research basis
- Dhuliawala et al., *Chain-of-Verification Reduces Hallucination in Large Language Models*: independent verification questions reduce anchoring to a draft in the evaluated tasks.
- Wu et al., *Large Language Models Can Self-Correct with Key Condition Verification*: reconstructing a decisive condition is more diagnostic than generic self-critique in the reported experiments.
- Local `evidence-reconciliation` and `deep-research` skills: authority, freshness, scope, directness, and contradiction govern source-backed conclusions.
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 |