feat(skills): add verifiable web research runner

[jienigoto]

Addresses #77.\n\nThis adds a canonical verifiable-web-research skill package with:\n\n- a deterministic CLI runner that turns captured source snapshots into claim-level evidence packets\n- exact quote validation so each claim must be backed by text present in the source content\n- audit-ready evidence archives, digest metadata, replay guidance, and optional JSON/Markdown artifacts\n- a fixture and JSON schema documenting the output contract\n- a Node built-in test covering successful artifact generation and invalid extract rejection\n\nValidation run locally:\n\n- node --test skills/verifiable-web-research/test.mjs\n- git diff --cached --check\n\nNote: I could not run the full pnpm workspace checks on this Windows machine because pnpm workspace install requires symlink creation and failed with EPERM. The new skill has a standalone test path that does not rely on pnpm or external network access.

[auscaster]

Thanks for pushing the skill PR. If this PR is for a Frantic bounty, the PR is the source/provenance trail; Frantic is the review and payment surface.

Next step: open the matching bounty on https://gofrantic.com/bounties, publish the skill with the current runx login path, then submit the delivery there with the required named artifacts:

npx --yes @runxhq/cli@latest login --provider github --for publish
npx --yes @runxhq/cli@latest registry publish ./skills/<skill-name>/SKILL.md --registry https://api.runx.ai

Include the live registry public_url, PR URL, raw SKILL.md, raw X.yaml, harness evidence, receipt reference, and report. Do not paste Connect session URLs, agent tokens, or payout secrets into GitHub.