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.
npx skills add skymanbp/CodeEraser --skill erasegit clone --depth 1 https://github.com/skymanbp/CodeEraserWrote 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.
[](https://agentmods.dev/skills/skymanbp/codeeraser/erase)<a href="https://agentmods.dev/skills/skymanbp/codeeraser/erase"><img src="https://agentmods.dev/badge/skills/skymanbp/codeeraser/erase/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/skymanbp/codeeraser/erase"><img src="https://agentmods.dev/badge/skills/skymanbp/codeeraser/erase.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 1 finding, up to high
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- high Anti-Refusal · line 29 Skill instructs the agent to omit warnings, disclaimers, or ethical commentary. Stripping safety caveats hides risk from the user and is a common jailbreak preamble.Fix: Remove instructions that suppress warnings, disclaimers, or ethical commentary. Let the agent surface safety-relevant caveats to the user.
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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00056 | $0.01048 |
| Opus 5.5 | $0.00022 | $0.00419 |
| Sonnet 5.5 | $0.00011 | $0.00210 |
| Haiku 4.5 | $0.00006 | $0.00105 |
Grade A, and why
erase 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 4d 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.
How it starts
The opening of the file, as written. The whole thing — 103 lines — stays where its author put it; the contents beside it link to each section on GitHub.
erase — from findings to deletions, safely
CodeEraser finds entropy; this skill is the OTHER half — acting on it without breaking anything. The contract: never delete what you have not read, and never claim convergence without re-running the gates.
0. Snapshot first
ce check . # the before-score and the current ratchet state
Record the score line. Every deletion below must leave ce check
passing; the score should not fall.
1. The deterministic pass FIRST — ce erase
ce erase . # dry-run plan: what is PROVABLY safe to remove
ce erase . --apply # act on it (git repo + clean worktree required)
ce erase . --log # the audit trail every apply appended (.ce/erase-log.ndjson)
Three classes erase without judgment (dead files, verbatim doc
duplicates, whole-unit byte-identical twins in dead files) — the tool
plans them deterministically, applies behind preconditions, and
proves its own convergence (contract: docs/reference/erase.md); the
trail it leaves is the same document the MCP tool erase_log and the
GUI erase screen's audit-log section show, and a record the reader
cannot parse is named by line. Do NOT hand-delete anything the plan
already covers. Everything it
prints as advisory is YOUR half — that is where the judgment below
begins.
2. Collect the advisory candidates (three signals, JSON faces)
ce dedup . --format json # T1/T2 clone blocks
ce deadcode . --format json # zero-liveness verdicts (graph-based)
ce join . --days 14 --format json # similarity × position × churn
ce join ranks the strongest deletion candidates: high similarity,
weak graph position, low churn. Start from the top of that list.
Four more families help where they apply (all advisory, the same documents the MCP tools and the GUI show):
ce merge . --format json # a clone group as ONE function: its parameters, the member kept, lines saved
ce flow . --format json # dead code INSIDE functions: unreachable statements, dead stores, unused locals
ce similar --unit <name/arity> . --format json # same-role twins of the unit you are about to keep
ce query 'dead(F)' . --format json # ask the index directly; `ce rules .` lists the rule violations
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.
- 4d ago Changed · +17 lines 8007c78f1f59
- 1mo ago Changed · +4 lines 6561fab03e3c
- 1mo ago First seen · 82 lines · 56 tokens per session scan A 2d0ee9a6d463
erase is a skill published in the GitHub repository skymanbp/CodeEraser (21 stars, last pushed today), licensed Apache-2.0. It adds 56 tokens to every session and 1,048 once invoked, about $0.0002 per session on Opus 5.5. 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 2026-08-30.
Other skills, from other repositories
health-check
Get an overall Python code quality health score using pyscn. Use when user asks how healthy or good the code is, wants a quality overview, a grade, a summary of technical debt, or a before/after quality comparison.
codeql-expert
Expert-level CodeQL for static analysis, vulnerability detection, and security code scanning. Use when the user mentions static analysis, SAST, vulnerability detection, or github security, or when the task involves writing CodeQL queries, building a CodeQL database, taint tracking, or scanning for injection and…
systematic-debug
A structured debugging process for investigating bugs, errors, failed tests, and unexpected behavior. It starts by creating a quick, repeatable way to reproduce the problem before testing possible causes.
audit-quality-gates
Use this skill when the user wants to audit how a project configures its style checkers and static analyzers.
scan
Scan a project with sentrux to get structural health grades (A-F) across 14 dimensions including coupling, cycles, cohesion, dead code, and test coverage. Use when the user wants to check code quality, architecture health, or before/after an agent session.
ubs
Ultimate Bug Scanner - Pre-commit static analysis for AI coding workflows. 13 languages, including bounded PHP request-flow analysis, with language-specific detection categories and native or ast-grep analysis. The AI agent's quality gate.