evolve-the-file

evolve-the-file is a skill for Claude Code from asyncswap/skills. It costs 53 tokens per session (781 once invoked), scanned A, original, MIT.

A test method for checking how a language server responds as a file changes through several complete snapshots. A language server is the program that powers editor features such as finding references and definitions.

In plain words
What is it for?
Use it to test editor requests after adding callers, moving code, or introducing new usages, with a separate cursor position for each version.
Why use it?
It catches errors caused by changed line positions, new code, or stale analysis between edits.

Skill for Claude Code

Written for Claude Code: shipped in a Claude Code plugin.

Part of the lsp-bench plugin — 12 skills shipped together

Good fit Use it to test editor requests after adding callers, moving code, or…

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/asyncswap/skills/evolve-the-file
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent
npx skills add asyncswap/skills --skill evolve-the-file
Clone the repo
git clone --depth 1 https://github.com/asyncswap/skills

Made for: Claude Code.

Or install lsp-bench, the plugin that ships this one along with the rest of its 12 skills.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for evolve-the-file

README.md
[![agentmods](https://agentmods.dev/badge/skills/asyncswap/skills/evolve-the-file.svg)](https://agentmods.dev/skills/asyncswap/skills/evolve-the-file)
Your own site
<a href="https://agentmods.dev/skills/asyncswap/skills/evolve-the-file"><img src="https://agentmods.dev/badge/skills/asyncswap/skills/evolve-the-file.svg" alt="Measured on agentmods" height="20"></a>
Per session 53 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 781 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00053 $0.00781
Opus 5 $0.00026 $0.00391
Sonnet 5 $0.00011 $0.00156
Haiku 4.5 $0.00005 $0.00078

Measured 6d ago against content hash f3052a801e75, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade A, and why

evolve-the-file scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 6d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

plugins/lsp-bench/skills/evolve-the-file/SKILL.md · 90 lines

How it starts

The opening of the file, as written. The whole thing — 90 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Sequence of edits with didChange:

didChange: sends snapshots of the file content via textDocument/didChange between iterations. Each step has its own (line, col) because edits shift positions.

Simple

methods:
  textDocument/references:
    line: 70
    col:  27
    didChange:
      - file: target.v2.snapshot
        line: 69
        col:  27
      - file: target.v3.snapshot
        line: 69
        col:  27

The bench runs the method N+1 times (baseline + one per snapshot). Each snapshot file holds the full edited content; the bench buffers it via didChange so the LSP recompiles before the next request.

Why per-step (line, col)

When the edit inserts code above your cursor, the original line shifts. The skill conventionally annotates each step with a comment explaining the shift:

methods:
  textDocument/definition:
    line: 137                # PRICE — immutable declared at line 68
    col:  32
    didChange:
      - file: Shop.v2.snapshot
        line: 142
        col:  32
        # PRICE shifted (getPrice added above)
      - file: Shop.v3.snapshot
        line: 138
        col:  32
        # PRICE same offset (event/error added at bottom — below cursor)
      - file: Shop.v4.snapshot
        line: 144
        col:  32
        # PRICE shifted (both edits combined)

Stable cursor (cursor above all edits)

If the cursor sits above every edit, positions stay constant — useful for checking response stability:

methods:
  textDocument/hover:
    line: 42
    col:  13
    # addTax — function in library section, above all edits
    didChange:
      - { file: target.v2.snapshot, line: 41, col: 13 }
      - { file: target.v3.snapshot, line: 41, col: 13 }
      - { file: target.v4.snapshot, line: 41, col: 13 }

(The line shifts by 1 because the snapshot file itself differs — adjust to match.)

Per-step expect:

Combine with expect: to assert how the response evolves:

methods:
  textDocument/references:
    expect: { minCount: 14 }                      # baseline
    didChange:
      - file: target.v2.snapshot
        expect: { minCount: 14 }                  # same — edit added no new refs
      - file: target.v3.snapshot
        expect: { minCount: 16 }                  # +2 new usages

Read the full file on GitHub · 90 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 6d ago First seen · 90 lines · 53 tokens per session scan A f3052a801e75

Subscribe to this mod's changes

evolve-the-file is a skill published in the GitHub repository asyncswap/skills (1 stars, last pushed 4mo ago), licensed MIT. It adds 53 tokens to every session and 781 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories

live-preview

Mid-build visual verification loop. Takes screenshots of components during construction, not just after. Catches visual regressions and invisible features before they compound. Requires Playwright or similar screenshot tool.

SethGammon/Citadel · 40 tokens

validate-album

Validates album directory structure, file locations, and content integrity. Use before release or whenever the user wants to check an album's structural health.

bitwize-music-studio/claude-ai-music-skills · 33 tokens

brooks-sweep

Full-sweep mode: runs a unified analysis across all quality dimensions — code decay, architecture, tech debt, and test quality — then applies fixes directly to the codebase. Safe changes are auto-applied; risky changes are confirmed before execution. Drawing on twelve classic engineering books. Triggers when: user…

hyhmrright/brooks-lint · 178 tokens

test-review

Test coverage review via Codex exec. Use when: reviewing test sufficiency, identifying coverage gaps, test quality audit. Not for: generating tests (use codex-test-gen), code review (use codex-code-review). Output: coverage analysis + gap report.

sd0xdev/sd0x-harness · 56 tokens

api-testing

HTTP API testing for TypeScript (Supertest) and Python (httpx, pytest). Test REST APIs, GraphQL, request/response validation, authentication, and error handling.

secondsky/claude-skills · 39 tokens

bun-jest-migration

Use when migrating from Jest to Bun's test runner, import compatibility, mocks, and config.

secondsky/claude-skills · 25 tokens