Proofread

Activation Boundary

This is a tag-only worker skill. Use it only after execute or execute-one selects an exact <@proofread...> annotation in a target document and hands off its context. Never activate it from an ordinary chat-window request, semantic similarity, or any skill other than those two routers. Without that router handoff, do not change the document; tell the user to add the tag and run one of the routers.

Fix grammar, spelling, and punctuation in the surrounding block, then consume the tag.

Tag

<@proofread>
<@proofread: optional narrower instruction>

Example: <@proofread: only fix punctuation>.

Input

• File path — the document being processed.
• Surrounding block — the maximal run of non-blank lines containing the tag; blank lines bound it.
• Prompt (optional) — narrows the scope. Default: fix all mechanical errors.

Workflow

1. Read the surrounding block; read the full file if broader context helps (for example, to recognize intentional terminology).
2. Fix grammar, spelling, and punctuation errors only.
3. Write the result back: replace the surrounding block (including the <@proofread> tag) with the corrected text.

Scope Modifiers

• Protect << >> — must NOT be modified even if it contains errors. Preserve the content exactly; remove the delimiters in the output.
• Field (( )) — ONLY text inside may be corrected. Remove the delimiters in the output.

Context References

Double-backtick references in the prompt point to resources to consult (for example a glossary of domain terms that must not be "corrected"). Resolve paths relative to the document's directory unless absolute. Remove the delimiters from the output.

Rules

• Do not change meaning, tone, or style; fix errors only, never rewrite.
• Consume this tag and its consumed modifier delimiters. Preserve every unrelated tag, <@comment: ...> tag, <@chat-model> block, and escaped literal byte-for-byte.
• On any error (for example an unreadable referenced file), fail cleanly: leave the tag, change nothing, report.