Skip to main content
skill CX Codex

indexion-identity — Claude Code skill: audit names in code

This skill checks whether function, file and folder names still match their contents, then plans verified renames. Useful during refactoring.

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add trkbt10/indexion-skills --skill indexion-identity -a claude-code
Open on GitHub

What indexion-identity does

The skill runs indexion identity audit over a project and turns the output into a review queue. It scans files, folders and graph-level declaration symbols. Local variables, parameters and fields are not treated as separate naming targets, though they can still shape the summary of the containing file. File names are evaluated with parent scope, and declaration-heavy files read as evidence-thin until follow-up inspection proves a real rename or split.

Every row is a review candidate, not proof. You compare name, expected_summary, actual_summary, assessment and recommendation. Before editing, the skill suggests confirming the finding with indexion doc graph --format=text <path>, indexion grep --semantic=name:<symbol> . and rg "<name-or-term>" <path>. Then you pick the smallest confirmed operation: rename a symbol, a file or a folder, move a file, or split it when several dominant responsibilities show up.

An insufficient-content assessment is an evidence problem, not naming drift. Check first whether the file is intentionally declarative, empty, or unsupported by the graph extractor. After changes the skill runs moon info && moon fmt, moon test and the audit again. Chasing a high score mechanically misses the point; verify actual code ownership and references first.

What it helps with

  • find functions whose name drifted from the implementation
  • check whether a file name still matches its declarations
  • decide if a file belongs in another folder
  • split a file with several dominant responsibilities
  • get a machine-readable JSON audit report

Example prompts for the agent

  • run the identity audit over the project and list rename candidates
  • check whether this file name is stale before I rename it
  • write the audit report to .indexion/cache/identity/report.json

How to install indexion-identity

1Skills CLI (Vercel)

npx skills add trkbt10/indexion-skills --skill indexion-identity -a claude-code

2Manually: put the files in place

~/.claude/skills/indexion-identity/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/indexion-identity/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about indexion-identity

How do I install indexion-identity in Claude?

Run npx skills add trkbt10/indexion-skills --skill indexion-identity -a claude-code. The skill lands in ~/.claude/skills/indexion-identity/SKILL.md.

What does indexion-identity do?

It scans function, file and folder names, compares them against their contents, and proposes operations to verify: renames, moves, folder changes or file splits.

What does the insufficient-content assessment mean?

It points to missing evidence rather than naming drift. Check whether the file is intentionally declarative, empty, or unsupported by the graph extractor before renaming anything.

Source text

SKILL.md · Apache-2.0
---
name: indexion-identity
description: Detect and review name/content drift in code using `indexion identity audit`, then plan verified renames, moves, folder changes, or splits.
---

# Identity Audit Workflow

Use this skill when checking whether function, file, or folder names still match
their contents.

The CLI scans files, folders, and graph-level declaration symbols. It does not
treat every local variable, parameter, or field as an independent naming target;
those can still influence the containing file summary. File names are evaluated
with parent scope, and declaration-heavy files should normally be read as
evidence-thin until follow-up inspection proves an actual rename or split.

## Pipeline

1. Run the mechanical scan:

   ```bash
   indexion identity audit .
   ```

   For a machine-readable queue:

   ```bash
   indexion identity audit --format=json --output=.indexion/cache/identity/report.json .
   ```

2. Treat each row as a review candidate, not proof. Compare:

   - `name`: the scoped name inferred by the identity package
   - `expected_summary`: what the path/name/scope predicts
   - `actual_summary`: graph-derived declarations, docs, module notes, and path terms
   - `assessment`: whether this is actual drift, overbroad content, or insufficient content
   - `recommendation`: first operation to verify

3. Verify before editing:

   ```bash
   indexion doc graph --format=text <path>
   indexion grep --semantic=name:<symbol> .
   rg "<name-or-term>" <path>
   ```

4. Choose the smallest confirmed operation:

   - Treat `insufficient-content` as an evidence problem, not a naming-drift proof.
     Inspect whether the file is intentionally declarative/thin, unsupported by the
     graph extractor, empty, or missing doc/declaration material before renaming.
   - Rename a symbol when its implementation is cohesive but the name is stale.
   - Rename a file when its declarations are cohesive but the file name is stale.
   - Rename a folder when contained files share a clearer parent concept.
   - Move a file when it fits an existing folder better than its current one.
   - Split a file when candidates show multiple dominant responsibilities.

5. After changes:

   ```bash
   moon info && moon fmt
   moon test
   indexion identity audit .
   ```

The audit is designed to surface review work. Do not maximize the score
mechanically; verify the actual code ownership and references first.

© trkbt10 · Apache-2.0 · trkbt10/indexion-skills

Similar skills

Everything for Code review and refactoring →

From the same set6

Project sponsors

Companies that keep this analytics open
Gold sponsor

Impressions, clicks and code copies — open statistics. We keep the numbers public for the same reason NeBlask itself is public. Become a sponsor