majiayu000/spellbook / skills/xray
xray
VerifiedInvestigate how a concept, code path, application, network flow, system, incident, document, or local artifact actually works, then deliver a two-depth visual HTML explainer with a dead-simple big-picture first layer and source-backed technical depth behind it. Use when the user invokes $xray, asks how something works under the hood, wants a module or app behavior traced across boundaries, requests web, code, network, or safe static artifact research before explanation, or asks for an evidence-backed explainer. Do not use for answer-only definitions, general web design, full codebase audits, or invasive binary reverse engineering without explicit authorization.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 13.2 KB |
| agents/openai.yaml | config | 379 B |
| assets/concept-explainer.html | other | 9.0 KB |
| assets/feature-explainer.html | other | 7.9 KB |
| evals/evals.json | config | 7.9 KB |
| references/chinese-writing.md | markdown | 3.8 KB |
| references/code-archaeology.md | markdown | 4.7 KB |
| references/evidence-contract.md | markdown | 3.2 KB |
| references/research-routing.md | markdown | 2.8 KB |
| references/reverse-core.md | markdown | 8.2 KB |
| references/visual-explanation.md | markdown | 5.1 KB |
| scripts/render-check.shexecutable | script | 1.9 KB |
references/evidence-contract.md
View on GitHub# Evidence Contract
The explainer may be simple; its evidence standard is not.
## Source Priority
For web research, prefer:
1. Standards, specifications, original papers, official documentation, source code, and first-party datasets.
2. Official release notes, engineering posts, issues, and pull requests.
3. High-quality secondary analysis that identifies its sources.
4. Community discussion as a lead or experience report.
5. Model output only as a search lead, never as evidence.
For repository work, prefer:
1. Current source, tests, schemas, configuration, runtime state, and generated contracts.
2. Current repository documentation.
3. Git history, issues, and pull requests for rationale.
4. Comments and names as hypotheses until verified by reachability or tests.
For incidents, prefer:
1. Persisted records, logs, metrics, request traces, and the running revision.
2. Configuration and source for that exact revision.
3. Design documents and expected behavior.
4. Browser memory or screenshots only when the UI itself is the disputed fact.
## Confidence Language
| Term | Meaning | Use |
|---|---|---|
| Observed | Directly present in the inspected artifact, source, runtime, or data | May be stated as fact with an anchor |
| Corroborated | Supported by independent authoritative evidence | May be stated as fact with sources |
| Inferred | Best causal explanation from observed evidence | Label as inference and show the reasoning |
| Unknown | Material evidence is missing or contradictory | Keep visible; do not fill the gap with a guess |
Use these terms only where uncertainty matters. Do not turn them into a badge system or require one status per paragraph. Agreement between models does not create corroboration because the outputs may share training data or repeat the same unsupported claim.
## Web Research Procedure
1. Search for the exact mechanism, specification, error, symbol, or version.
2. Open the authoritative page rather than citing a search result snippet.
3. Record publication or update dates and applicable versions where they affect the claim.
4. Corroborate behavior-changing or disputed claims with an independent primary source when possible.
5. Save the direct URL and a short note about the supported claim.
6. If sources conflict, identify whether they describe different versions, scopes, defaults, or implementations.
## Research Notes
Keep lightweight notes for pivotal claims before drawing. A compact form may be useful:
```text
claim: The retry happens before provider submission.
evidence: validation returns before submit_request is called.
anchor: services/gateway/request.go :: validateAndSubmit
status: observed
visual: validation gate before the provider boundary
```
This is a thinking aid, not a required schema or deliverable. Use another note shape when it fits the investigation better.
## Failure Rules
- Missing source for a pivotal claim: remove the claim or label it unknown.
- Broken or inaccessible source: find an authoritative replacement or disclose the limitation.
- Documentation disagrees with code: show the difference and identify which version is running.
- Runtime disagrees with local code: resolve the deployed revision before explaining the cause.
- Example not tested or observed: label it illustrative, not evidence.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |