TheGoat395/Codex-Skills / skills/web-reference-research
web-reference-research
VerifiedAnalyze selected sources: Viktor or MotionSites.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 1.9 KB |
| agents/openai.yaml | config | 249 B |
| references/evidence-record-schema.md | markdown | 2.9 KB |
| references/reconstruction-manifest.md | markdown | 921 B |
| references/web-reference-research-guide.md | markdown | 3.8 KB |
| scripts/verify-reference-record.py | script | 6.7 KB |
references/evidence-record-schema.md
View on GitHub# Structural evidence record schema
Each required field is a typed record: `{"status":"pass","evidence":[{"kind":"note","value":"Concrete check, result and scope"}]}`. Evidence kinds: `note`, `file`, `url`, `command`; values must be nonblank strings. Evidence paths are references only: this checker does not open or certify them. Before claiming file-backed proof, check existence and inspect contents within already authorized source roots; apply exclusions and symlink boundaries first.
Statuses: `pass`, `fail`, `untested`, `not-applicable`, `observed`, `inferred`. Fail, untested, N/A and inferred need a nonblank `rationale`. Pass/observed/inferred need nonempty evidence. A failed finding can contain a truthful negative test result. N/A has an empty evidence list if appropriate and describes why the requirement does not apply; it is never counted as a pass. Use observed/inferred for research rather than implying a runtime test.
`--require-passed` rejects declared fail, untested or inferred records; even a successful invocation only establishes structure. JSON cannot establish the truth of its own claims, rendered inspection or production certification.
Required fields: `source_url`, `creator`, `observed_on`, `evidence`, `sampled_states`, `hierarchy`, `palette`, `typography`, `layout`, `motion_phases`, `input_behavior`, `responsive_evidence`, `reduced_motion`, `technical_inference`, `confidence`, `performance_risk`, `clean_room_brief`, `originality_delta`, `unavailable_evidence`.
Fields permitting a truthful empty list: `unavailable_evidence`. Other fields require the record envelope. This explicitly replaces the earlier presence-only format; migrate old records rather than treating legacy strings/booleans as verified evidence.
## Version and compatibility
New manifests use `{"schema_version":2,"skill":"web-reference-research","records":{...}}`. Each record additionally includes `applicable: true` for applicable work or `false` for justified `not-applicable`; `null` is allowed only with `untested` while applicability is unknown. Existing unversioned typed records remain accepted as explicitly labeled compatibility input, not inferred legacy truth.
To preserve an old presence-only manifest, run the helper with `<old.json> --migrate-legacy --dry-run` to inspect the conversion or `<old.json> --migrate-legacy --output <new.json>` to create a new file exclusively. The source and an existing destination are never overwritten. The v2 envelope retains the entire original object and source SHA-256 under `legacy_source`, plus each required field's original value. Every migrated requirement is `untested` with unknown applicability; no string, list, false boolean, or other old truthy value is converted to `pass`. Review actual evidence before adopting a status. Migration success describes file creation only; validation of a migrated manifest with unresolved records returns non-pass. Dry-run output is a preview, not saved evidence.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| add-product-e2e-evalpaperclipai/paperclipAdd or extend a Paperclip full-stack runner E2E workflow, fixture, matcher, or report evidence path for local or Daytona execution. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| add-runner-evalpaperclipai/paperclipAdd or extend a Paperclip Runner protocol evaluation definition, roster, assertion, or report fixture with provenance and narrow validation. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| diagnose-why-work-stoppedpaperclipai/paperclipCompatibility entry point for failed, stalled, looping, or over-recovered Paperclip work. Use fix-failed-paperclip-run for the full diagnosis, regression, repair, real verification, and green-PR workflow; honor an explicitly requested diagnosis-only scope. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| fix-failed-paperclip-runpaperclipai/paperclipDiagnose and fix failed, stuck, slow, or incorrectly blocked Paperclip tasks and runs. Use with a task/run URL or failure report to trace the root cause, prove a regression red then green, check second-order effects, verify the real workflow, and deliver a PR with passing checks. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |