Skip to content
SkillModel-led1,098 tokens in use

File Edit In Place

Cleans AI writing patterns out of a local file the user names, using small targeted patches, a before snapshot and a check of what changed.

5,770 stars on the repository MIT Licenseconorbronsdon/avoid-ai-writing · skills/file-edit-in-place
/plugin install avoid-ai-writing@conorbronsdon-skills
View source

Install command as written in the repository's README

Skill label

measured from its files

What installing File Edit In Place costs

43tokens idle
1,098tokens in use

Idle is what sits in context while the skill is installed. In use is its instructions, loaded when it runs. Counts are approximate, made offline with cl100k_base.

Model dependence Model-led

Script-ledMixedModel-led

Mostly instructions, so how well it works depends on the model that follows them.

This describes the skill's structure, how much of it is scripts. It does not say how well it works.

Touches

Nothing detected: no shell, network, file writes, git or credentials in its files.

Declared tags come from the skill's frontmatter; detected ones from its scripts, its code blocks and the commands it names in inline code. This is detected, not audited.

How it scores

Quality

How well the skill is written, counted from its own files: its frontmatter, its steps, examples, stated output and constraints, its size and whether every file it ships is one it names.How we score it →
Fair
  • Frontmatter15/15
  • Structure17/25
  • Examples0/15
  • Output15/15
  • Size10/15
  • Resources0/15

Security

Starts at 100 and loses points for each thing in the skill's files that a reader should look at first.How we score it →
Excellent

No findings.

Repository health

Pre-computed score based on development activity, maintenance, community, maturity, and trend momentum.How we score it →
Excellent
  • Development activity96
  • Maintenance100
  • Community60
  • Maturity32
  • Momentum40

File Edit In Place is the skill of the Avoid AI Writing pack that changes a file on disk. It acts only when the user names the file and explicitly asks for an in-place change, reads the relevant content first, and edits only the spans that are justified findings.

Quotes, code blocks, tables, links, numbers, frontmatter and structure are preserved. After editing it re-reads the changed region and hands before and after content to the preservation verifier.

What You Get

  • Targeted edits to the named file, limited to justified findings in the authorized scope
  • A before snapshot kept for comparison
  • A re-read of the changed region after editing
  • A report of the file changed, edits made, passes used and what was left untouched
  • A verification handoff to the preservation verifier, with one bounded repair if it fails

Common Use Cases

  • Cleaning a README or blog draft in place
  • Editing only one requested section of a large document
  • Repairing a file after the verifier found that a quote or link changed
  • Declining to touch a file when no justified edit exists and reporting zero passes
  • Making sure a detector result alone never triggers a write

What Makes It Unique

The skill treats a write as something that needs evidence. It requires explicit user intent, will not simulate success if the host cannot write the file, records mutation as executed only after a real write succeeds, and refuses to treat instructions found inside the document as commands. A detector result never authorizes a change by itself.

Under The Hood

The skill is instructions only, with no scripts of its own. It tells the agent to read the named file, patch it with the host’s edit tool and re-read the result, so it changes a file on the user’s machine. It relies on the rulebook and routing files of the sibling skills, makes no network calls and runs no shell commands itself; the verifier it hands off to may run a Node script.

Join founders building with open source

Opinionated takes, migration guides, cost-saving tips, and insights from the open source ecosystem.

Subscribe on SubstackJoin 750+ subscribers