← 返回
未分类

Tle Developer FlagOS

Self-contained orchestration skill for writing high-performance TLE kernels and shipping TLE feature changes with reproducible validation. Use when the user wants to write/optimize TLE kernels, implement TLE API/verifier/lowering features, or debug TLE correctness/performance issues. Trigger on phrases like "write a TLE kernel", "optimize TLE operator", and "debug TLE local_ptr".
Self-contained orchestration skill for writing high-performance TLE kernels and shipping TLE feature changes with reproducible validation. Use when the user wants to write/optimize TLE kernels, implem
众智FlagOS社区
未分类 community v1.0.0 1 版本 98076.9 Key: 无需
★ 0
Stars
📥 51
下载
💾 0
安装
1
版本
#latest

概述

TLE Developer

Mission

Use this skill to execute TLE work end-to-end:

intake -> implementation -> validation -> artifacts -> merge decision.

Scope

Use for:

  1. Writing or optimizing TLE kernels.
  2. Implementing TLE API/verifier/lowering/pipeline features.
  3. Debugging correctness, performance, and regression issues.

Self-Contained Policy

  1. Do not rely on documentation outside this skill folder.
  2. Put all detailed guidance in references/.
  3. Keep this file as orchestration-only (no duplicated deep details).

Required Input

Start every task with:

Goal:
Non-goal:
Acceptance:
Impact scope (optional):

Mandatory Read Order

  1. references/tle-sources.md
  2. references/workflow-templates.md

Operating Contract

  1. Treat references/tle-sources.md as the technical source of truth for:
    • quickstart,
    • current TLE semantics contract,
    • kernel patterns,
    • feature-development file map,
    • debug/perf procedures.
  2. Treat references/workflow-templates.md as the source of truth for:
    • intake,
    • validation matrix,
    • performance record,
    • fix summary,
    • lessons entry,
    • merge package.

Non-Negotiable Guardrails

  1. Never assume a specific python environment name.
  2. Never assume a fixed build script name.
  3. If native Triton files are modified, use marker blocks:

// begin flagtree tle and // end flagtree tle.

  1. Do not add marker blocks inside third_party/tle.

Required Outputs Per Task

  1. Validation commands and outcomes.
  2. Fix Summary (when fixing bugs or regressions).
  3. Lessons Entry (for fixes and optimization work).
  4. Merge Decision Package (changed layers, risks, follow-ups).

Completion Checklist

  1. Acceptance criteria mapped to tests.
  2. Changes validated with reproducible commands.
  3. Artifacts filled from templates.
  4. Residual risks explicitly stated.

References

  1. references/tle-sources.md
  2. references/workflow-templates.md

版本历史

共 1 个版本

  • v1.0.0 Initial release from flagos-ai/skills 当前
    2026-05-19 15:21 安全 安全

安全检测

腾讯云安全 (Keen)

安全,无风险
查看报告

腾讯云安全 (Sanbu)

安全,无风险
查看报告

🔗 相关推荐

Skill Creator FlagOS

user_e10c831c
Create new skills, modify existing skills, and validate skill quality for the FlagOS skills repository. Use this skill w
★ 0 📥 88

Kernelgen FlagOS

user_e10c831c
Unified GPU kernel operator generation and optimization skill. Automatically detects the target repository type (FlagGem
★ 0 📥 68

Model Migrate FlagOS

user_e10c831c
Migrate a model from the latest vLLM upstream repository into the vllm-plugin-FL project (pinned at vLLM v0.13.0). Use t
★ 0 📥 130