Agent Skillsmohitagw15856/pm-claude-skills › form-filler-operator

form-filler-operator

GitHub

通过计算机使用代理,基于用户确认的事实表逐项填写长网页表单。生成事实映射、暂存待审的已填表单及字段级转录记录,严禁猜测数据或自动提交,敏感信息需人工输入。

skills/form-filler-operator/SKILL.md mohitagw15856/pm-claude-skills

Trigger Scenarios

帮我填写这份申请 完成政府/供应商/保险表格 进行此注册

Install

npx skills add mohitagw15856/pm-claude-skills --skill form-filler-operator -g -y
More Options

Use without installing

npx skills use mohitagw15856/pm-claude-skills@form-filler-operator

指定 Agent (Claude Code)

npx skills add mohitagw15856/pm-claude-skills --skill form-filler-operator -a claude-code -g -y

安装 repo 全部 skill

npx skills add mohitagw15856/pm-claude-skills --all -g -y

预览 repo 内 skill

npx skills add mohitagw15856/pm-claude-skills --list

SKILL.md

Frontmatter
{
    "name": "form-filler-operator",
    "description": "Fill long web forms and applications through a computer-use agent — from a fact sheet you approve, field by field, with a full transcript and nothing submitted without your word. Use when asked to fill this application for me, complete this government\/vendor\/insurance form, or do this registration. Produces the fact-to-field mapping, the filled form held at review, and a field-level transcript."
}

Form Filler Operator Skill

Forms are where an hour of a professional's day goes to die: the same facts, retyped into someone else's boxes. This skill runs the retyping through a computer-use agent — with the discipline the task actually needs: facts come from an approved sheet (never invented), every field write is logged, and the submit button belongs to the human. (Reading the form's fine print first? That's tos-decoder's job — chain them.)

What This Skill Produces

  • The fact sheet — every fact the form needs, gathered once, confirmed before any filling
  • The mapping — form field → fact → source, including fields with no matching fact (asked, never guessed)
  • The filled form, held at review — completed up to (never through) the submission step
  • The transcript — field-by-field record of what was entered where

Required Inputs

Ask for these if not provided:

  • The form — URL or document, and any login the user has already established
  • The facts — documents/profile to draw from (or interview the user to build the sheet)
  • Sensitivity rules — which fields (SSN/passport/banking) require the user to type them personally; default: ALL such fields are user-typed
  • The deadline and stakes — a visa application and a newsletter signup deserve different paranoia

Framework

  1. Sheet before form: all facts assembled and user-confirmed first; mid-form invention is how wrong data gets submitted.
  2. Map, then fill: every field mapped to a fact + source; unmapped fields become questions batched to the user — one interruption, not twenty.
  3. Sensitive-field floor: identity, financial, and medical fields are never agent-typed unless the user explicitly overrides, field by field.
  4. Checkpoint pages: multi-page forms get a per-page pause-and-verify against the mapping before proceeding.
  5. The held-at-review rule: the run ends at the review/submit screen with a screenshot; pressing submit is the user's hand or an explicit word-level go.

Output Format

Form Run: [form name] — [date]

Fact sheet: [n] facts confirmed · Fields: [n] mapped, [n] user-typed (sensitive), [n] open questions

Field Entered Fact source
Status: held at review — [screenshot] — awaiting your submit.

Quality Checks

  • Zero fields filled from inference — every entry traces to the confirmed sheet
  • All sensitive fields were user-typed or explicitly overridden one by one
  • Open questions were batched, not guessed
  • The run ended held-at-review with visual proof

Anti-Patterns

  • Do not guess a field to keep momentum — a wrong passport number costs weeks
  • Do not accept ToS/declarations on the user's behalf — decode them (tos-decoder) and put the checkbox in the user's hands
  • Do not autofill from a browser profile — the confirmed sheet is the only source
  • Do not treat CAPTCHAs or identity checks as obstacles to defeat — they're the user's to complete, by design

Execution

For computer-use agents with browser control. Without tools, the fact sheet + mapping is the deliverable (still worth an hour). Rules per SKILLSPEC.md §5.

Preconditions

  • The fact sheet confirmed by the user; the mapping approved; sensitivity rules acknowledged.
  • The user is present (or reachable) for sensitive fields, CAPTCHAs, and the final submit.
  • For logged-in forms: the session was established by the user, not the agent.

Allowed actions

  • Navigate the named form; enter mapped facts into their fields; upload documents the user staged for this run.
  • Pause at each page checkpoint and at every sensitive field; screenshot at each checkpoint.
  • Stop at the review/submit screen.
  • Nothing else: no submitting, no account creation, no payment entry, no accepting agreements, no navigating beyond the named form.

Verification

  • At review: read back every field from the form itself against the mapping; list any field the form transformed (formatting, truncation).
  • Deliver the transcript + review screenshot.

Rollback

  • Unsubmitted forms roll back by closing the session (nothing was committed).
  • Stop and ask a human if: a field rejects a mapped fact, the form's structure changes mid-run, an unexpected identity/payment step appears, or anything requires agreement acceptance.

Version History

  • 54fad50 Current 2026-07-19 13:28

Same Skill Collection

exports/openclaw/360-feedback-template/SKILL.md
exports/openclaw/401k-plan-decoder/SKILL.md
exports/openclaw/ab-test-planner/SKILL.md
exports/openclaw/ab-test-readout/SKILL.md
exports/openclaw/accessibility-audit/SKILL.md
exports/openclaw/account-plan/SKILL.md
exports/openclaw/acquirer-red-team/SKILL.md
exports/openclaw/ad-copy/SKILL.md
exports/openclaw/aeo-optimizer/SKILL.md
exports/openclaw/agenda-or-cancel/SKILL.md
exports/openclaw/agent-design-review/SKILL.md
exports/openclaw/agent-observability-spec/SKILL.md
exports/openclaw/agent-spec/SKILL.md
exports/openclaw/ai-ethics-review/SKILL.md
exports/openclaw/ai-eval-plan/SKILL.md
exports/openclaw/ai-feature-prd/SKILL.md
exports/openclaw/ai-product-canvas/SKILL.md
exports/openclaw/air-quality/SKILL.md
exports/openclaw/altitude-shifter/SKILL.md
exports/openclaw/ambiguity-resolver/SKILL.md
exports/openclaw/analyst-relations-brief/SKILL.md
exports/openclaw/announcement-card/SKILL.md
exports/openclaw/api-docs-writer/SKILL.md
exports/openclaw/api-test-plan/SKILL.md
exports/openclaw/api-versioning-strategy/SKILL.md
exports/openclaw/apology-letter/SKILL.md
exports/openclaw/architecture-decision-record/SKILL.md
exports/openclaw/architecture-diagram/SKILL.md
exports/openclaw/archive-strategy/SKILL.md
exports/openclaw/assumption-bounty/SKILL.md
exports/openclaw/assumption-mapper/SKILL.md
exports/openclaw/async-update-format/SKILL.md
exports/openclaw/auto-repair-estimate-decoder/SKILL.md
exports/openclaw/autopilot-charter/SKILL.md
exports/openclaw/benefits-decoder/SKILL.md
exports/openclaw/bid-tender-review/SKILL.md
exports/openclaw/board-deck-narrative/SKILL.md
exports/openclaw/board-minutes/SKILL.md
exports/openclaw/board-pre-read/SKILL.md
exports/openclaw/bom-cost-review/SKILL.md
exports/openclaw/bookkeeping-categorization/SKILL.md
exports/openclaw/boolean-search-builder/SKILL.md
exports/openclaw/brag-doc/SKILL.md
exports/openclaw/brainstorming/SKILL.md
exports/openclaw/brief-builder/SKILL.md
exports/openclaw/briefing-note/SKILL.md
exports/openclaw/budget-builder/SKILL.md
exports/openclaw/budget-variance-analysis/SKILL.md
exports/openclaw/bug-diagnosis/SKILL.md
exports/openclaw/bug-report/SKILL.md

Metadata

Files
0
Version
471c606
Hash
440edf4b
Indexed
2026-07-19 13:28

- 위키
Copyright © 2011-2026 iteam. Current version is 2.155.2. UTC+08:00, 2026-07-30 12:24
浙ICP备14020137号-1 $방문자$