Spellcheck Proofread & Polish
Content Creator Official v1.0.1Flags misspellings with a deterministic zero-token builtin, then an LLM applies those fixes and polishes grammar and voice — so the model never pays tokens to check whether something is a word.
by skrptiq
Sign in to import this skrpt into your workspace.
Sign inWhat's included 3 nodes
Spellcheck Proofread & Polish
Flags misspellings deterministically with a zero-token builtin, then an LLM applies the fixes and polishes grammar and voice.
Proofread Polish
Applies the spellchecker's flagged fixes and polishes grammar and voice, without re-scanning for spelling.
Polish Draft
Applies the flagged spelling fixes and polishes grammar and voice.
Spellcheck
built-in · offline · 0 tokensFlag misspelled words against an embedded dictionary — a list, not a rewrite.
Requirements 1 service
- LLM Service
Security Passed
Detected
- Services: llm-service
- Permissions:
- Data Handling:
Version history 2 releases
GH#820 — migrate the `spellcheck` builtin to the node-less `local.builtin` shape (#802 D1b): `id: spellcheck` + `input: {{input.draft}}` + `context`, no backing skill node/edge; downstream ref `{{steps.Spellcheck.output}}` → `{{steps.spellcheck.output}}`. Deterministic output unchanged. contents skills 2→1.
Initial release. The 5th #808 builtin-showcase skrpt. Showcases the flagship hybrid pattern: the deterministic `spellcheck` builtin flags misspellings (zero-token, offline, `en`) → a single LLM pass applies those fixes and polishes grammar and voice, without paying tokens to detect spelling. Deterministic assertion: the fixture `"We recieve teh report and it occured twice."` yields `flaggedCount: 3` (recieve, teh, occured).
More from Content Creator
View all →Profile-Aware Social Media Pipeline
Generate personalized multi-platform social content using your voice and audience profiles
Content Approval Pipeline
Draft content, pause for human review, then revise and polish based on your feedback
Cross-Provider Technical Review
Draft with one AI model, critique with another, revise until the reviewer passes — better output through model diversity