- Home
- Skills
- Frontend & Web Apps
- Form Validation Audit
Works with the AI Agents you already use
Form Validation Audit
Find confusing form errors, lost input and unsafe server feedback. Get evidence-linked fixes and a practical validation/recovery acceptance plan.
$9
Form Validation Audit
Example session with this skill installed
Fictional source-only audit example. Source-only profile form: email and display name editable. Contract requires localized field errors after submit and form-level permission/provider errors. Email described-by points at hint only. Invalid email shows red border and generic toast. Every edit clears all errors. Failure resets values and shows raw provider body. Runtime/focus/screen-reader evidence absent.
- Read your context and instructions
- Compiled the form validation audit
Profile form validation audit
Scope and evidence
Provisional supplied-source review for an actor editing contact email and display name to save a profile. Environment, revision, locale, actual display-name constraints and helper ownership are unspecified. The declared contract requires localized inline errors after submit; permission and provider outages remain form-level. No implementation or external actions are authorized.
- E1: Email
aria-describedbyreferences an existing hint only; invalid email receives a red border and generic toast. - E2:
onChangeclears the entire error map. - E3: Failure resets both values and displays the raw provider body.
- C1: Supplied validation and server-failure contract above.
These are source descriptions without file/symbol identifiers. Runtime, focus, screen-reader, deployed-revision, existing-helper and save/persistence evidence are absent. Display name is editable; no extra required, length or format rule is inferred.
Findings
| Finding | Evidence and consequence | Severity / confidence | Rule and narrow remedy | Reproducible check |
|---|---|---|---|---|
| Confirmed source defect: safe entries lost on failure | E3 resets email and display name, making users repeat work | High / high for supplied source; deployed outcome unknown | Preserve safe entered values on failed save | Inject a controlled failure with fictional entries; assert exact entries remain |
| Confirmed source defect: unsafe provider feedback | E3 displays unfiltered provider body; potential sensitive diagnostic exposure, actual body contents unknown | High / high for rendering behavior, unknown for actual exposure | C1: map documented failures to safe localized form-level messages; exclude raw bodies from UI and audit artifacts | Controlled provider/permission responses show safe non-field feedback with no raw diagnostic text |
| Confirmed contract defect: email feedback lacks specific inline guidance | E1 offers color and generic toast; existing described-by points only to hint | Medium / high | C1: add localized email-format error beside email using established helpers; retain hint association and expose the custom invalid state/error association | Invalid email after submit shows specific adjacent message; inspect valid IDs and supported association/invalid state |
| Confirmed lifecycle defect: unrelated issues erased | E2 clears every error whenever either field changes | Medium / high; simultaneous field issues depend on actual rules | Clear/revalidate only the edited field and real dependencies | With two independently supported issues, edit email; unresolved display-name issue remains. If display name has no independent rule, use an appropriate controlled lifecycle test without inventing a domain rule |
Focus and announcements are unverified, not confirmed missing. A toast can supplement inline feedback. Permission/outage messages belong at form level; inventing a field issue for them would be incorrect.
Before / after and acceptance plan
First preserve values and replace raw bodies; next repair email guidance and scoped clearing. Reuse the current validation/API owners and visual identity. No form-library replacement is justified.
| Scenario | Proposed acceptance evidence | Status |
|---|---|---|
| Invalid submit | Component and authorized browser: localized inline email error after submit; existing hint retained; predictable keyboard recovery to first invalid field or established actionable summary | Open |
| Correction | Component/browser: correction removes or revalidates email issue; unrelated current issues remain; entered values retained | Open |
| Permission/provider failure | Controlled fixture then authorized browser: safe localized form-level feedback and preserved entries; correction/retry possible | Open |
| Successful correction/save | Authorized browser: explicit confirmation, reload retains saved profile, every affected profile display reflects it | Open |
| Assistive/responsive use | Actual screen reader plus keyboard, supported locale and zoom: error understandable, associated and reachable | Open |
No tests, browser or assistive checks were run. Next, frontend and API owners should identify the existing helper and safe error contracts, then implement and verify these scoped checks in an authorized fictional-data environment. This limited audit does not certify accessibility or security.
All example data is fictional. No real form submissions or runtime checks were performed.
Connects securely to your tools. The creator never sees your data.
About this skill
Help your agent audit the states users actually encounter: invalid submit, field correction, async responses, server failure and successful save. Check feedback, programmatic error associations, keyboard recovery, safe retained values and downstream persistence. Preserve working native validation and existing form patterns instead of requiring a wholesale rewrite.
Includes an audit-report template, acceptance matrix, audit-method reference and fictional before/after example. Findings distinguish confirmed defects from risks and unverified runtime behavior. Actual browser and screen-reader verification require compatible host tools and an authorized test environment. This is a focused validation workflow, not an exhaustive accessibility/security certification or automatic form-fixing tool.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Frontend & Web Apps.
Focused validation review only. Evidence and source access limit findings. Three fictional Codex scenarios were checked; actual browser/screen-reader combinations and other agents were not tested. No full accessibility/security certification, guaranteed defect detection or automatic form fixes/submissions.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 22 days ago
Permissions
Reads bundled Markdown and user-supplied sanitized form evidence. Source/browser/assistive testing is optional and requires the host tools and task authorization. No real form submissions, saved production records, messages, payments, edits or deployments are automatically authorized.