Add Guardana for LLM01 and ASI01 testing

[karauda]

What this PR changes

Adds Guardana to the tool list and to the LLM01 / ASI01 sections of the MITRE ATLAS mappings. I maintain Guardana; it's an Apache-2.0 CLI in beta.

The rule catalog documents its direct and indirect prompt-injection checks. The two entry JSON files and the webapp bundle were regenerated from the mappings.

Type of change

• New mapping file
• Update to existing mapping (content, controls, CVE refs)
• Bug fix (broken link, typo, incorrect cross-ref)
• New recipe (shared/RECIPES.md)
• New tool (shared/TOOLS.md)
• Infrastructure (scripts, CI, templates)
• Translation (i18n/)

Source / evidence

• Rule catalog at 1.0.0rc2: guardana.prompt.injection.ignore_previous, guardana.scenario.indirect_injection and guardana.agent.tool_result_injection, with LLM01:2026 / ASI01:2026 mappings.
• Detection limits.

Checklist

Content

• Existing file structure retained; tools, references and changelogs updated.
• Severity ratings and technique mappings unchanged.
• No new cross-references between framework files.
• Referenced vulnerability IDs are valid: LLM01 and ASI01.
• Existing CC BY-SA 4.0 headers retained.

Links & data

• Internal links pass validation.
• New external URLs return 200.
• Generated entries remain compatible with the data schema.

Project hygiene

• Changelog entry added to every edited Markdown file.
• No new mapping file; root CHANGELOG.md and README mapping counts need no change.
• Tool-category count updated in shared/TOOLS.md.
• Both edited mapping files pass node scripts/validate.js --file <path>.

For new mapping files only

Not applicable: this PR adds no mapping file, framework or control.

Notes for reviewers

npm run build passes. npm test passes all 137 tests, with none skipped. The validator reports 92 warnings; I ran it on the unchanged base as well and the warning list is identical. Both edited mappings validate with zero errors and zero warnings.