Daniel AvilaCreated claude-code-templates, 32k+ GitHub stars346 installs on skills.sh

Speculative Decoding

Accelerate LLM inference using speculative decoding.

Works in any agentClaude CodeCodexCursorClaude appChatGPT+ any other agent
davila7/claude-code-templates
Security state: reviewed

3 static scans, 0 counted findings.

Edge checks it again each time your agent loads it.

SKILL.md3 files · 32.0 KB

From the davila7 public repository. Edge did not write or edit this skill.

What it does

  • Accelerate LLM inference using speculative decoding, Medusa multiple heads, and lookahead decoding techniques.
  • Use when optimizing inference speed (1.5-3.6× speedup), reducing latency for real-time applications, or deploying models with limited compute.
  • Covers draft models, tree-based attention, Jacobi iteration, parallel token generation, and production deployment strategies.
DA

Why Daniel Avila

Co-founder of CodeGPT. Each fact links its public source.

  • 32k+ GitHub stars on claude-code-templates

    His CLI for configuring and monitoring Claude Code has over 32,000 stars.

    GitHub
  • Co-founder of CodeGPT, 2M+ downloads

    He co-founded CodeGPT, which has over 2 million downloads.

    Personal site

Public sources, checked Oct 26, 2026.

How it works

  1. 1
    Copy

    Try for free gives you one prompt.

  2. 2
    Paste

    Paste it into Claude Code, Codex or Cursor.

  3. 3
    Ask

    Ask for the outcome. Your agent loads Speculative Decoding.

What you can inspect

3 static scans, 0 counted findings.

Security state at 2026-09-30: reviewed. Static scans check the code, not how well the skill works.

State definitions

Skill details

Security scansReviewed

3 static scans, 0 counted findings. Security state at 2026-09-30: reviewed. Edge checks it again each time your agent loads it.

Edge static checks (Edge-run): edge-static/1.0.1. Scope: Static patterns for remote execution, credentials with network sends, obfuscation, prompt overrides, hidden Unicode, binaries and persistence. Raw findings: .

Cisco skill-scanner (Edge-run): cisco-skill-scanner/2.1.0. Scope: Local static, YARA, pipeline and behavioral analyzers; no LLM or AI Defense analysis. Raw findings: 2.

Semgrep Edge rules (Edge-run): semgrep/1.178.0+edge-rules.c15016a36fd3+offline-v1. Scope: Six Edge-authored static rules with Semgrep CE; offline execution and metrics disabled. Semgrep-maintained rules excluded. Raw findings: .

Edge static checks
Pass · 0 counted findings · 2026-09-27
Cisco skill-scanner
Pass · 0 counted findings · 2026-09-27
Semgrep Edge rules
Pass · 0 counted findings · 2026-09-27
Creator, source & licencedavila7

From the davila7 public repository. Edge did not write or edit this skill.

Revision
6af7cdf57b810aca5e544a0f695bf058f1c2a744
Package hash
sha256:868b44e07dfd683a4993ad19a1429d5302012c11e0cb3784efae58bb7fa1e728
Licence
MIT
Description
Accelerate LLM inference using speculative decoding, Medusa multiple heads, and lookahead decoding techniques. Use when optimizing inference speed (1.5-3.6× speedup), reducing latency for real-time applications, or deploying models with limited compute. Covers draft models, tree-based attention, Jacobi iteration, parallel token generation, and production deployment strategies.
Quoted from the skill’s own description
Read skill files3 files

3 files, 32.0 KB at revision 6af7cdf57b81.

Questions about Speculative Decoding

What does Speculative Decoding do?

Accelerate LLM inference using speculative decoding. Published by Daniel Avila (davila7/claude-code-templates).

Is Speculative Decoding safe to use?

Edge security state: Reviewed (3 Edge-run static scans, scanned Sep 30, 2026). The Evidence tab shows the scan.

How do I use Speculative Decoding in Claude Code, Codex or Cursor?

Click Try for free, pick your agent and follow its one step. Edge is free, with no Edge account.

Similar skills

grill-with-docs
mattpocock

A relentless interview to sharpen a plan or design, which also creates docs (ADR's and glossary) as we go.

+31K installs this weekDocs
paper-context-resolver
lllllllama

Rigor Paper Context helper for README-first deep learning repo reproduction. Use only when the README and repository files leave a narrow re…

+6.7K installs this weekDocs
lark-doc
larksuite

飞书云文档(Docx / Wiki)内容操作:读取、创建、编辑文档,插入或下载图片附件,以及操作思维笔记。用户提供文档 URL/token(包括 doubao.com 的 /docx/、/wiki/)时使用;按 URL 路径/token 而非域名路由;`/page/<token>…

+5.6K installs this weekDocs
executing-plans
obra

Use when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or…

+6.9K installs this weekDocs
spec-driven-development
addyosmani

Creates specs before coding. Use when starting a new project, feature, or significant change and no specification exists yet. Use when draft…

+4.6K installs this weekDocs
gws-docs
googleworkspace

Read and write Google Docs.

+910 installs this weekDocs
golang-documentation
samber

Comprehensive documentation guide for Golang projects, covering godoc comments, README, CONTRIBUTING, CHANGELOG, Go Playground, Example test…

+409 installs this weekDocs
negotiation-concession-planner
mbfinotti

Builds a pre-negotiation concession plan for a pricing or contract conversation - every tradeable lever priced by cost-to-us vs value-to-the…

Docs