원클릭으로
skill-writing-guide
Use when creating, reviewing, or refining Agent Skills for this repository
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
메뉴
Use when creating, reviewing, or refining Agent Skills for this repository
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Use when debugging Cortex-R5 or Cortex-R firmware — TCM, MPU, caches, DFSR/DFAR aborts, exceptions, GIC/VIC interrupts, lockstep or split mode, ECC, or JTAG bring-up
Use when integrating or debugging heatshrink embedded compression — encoder/decoder sink/poll/finish loops, window/lookahead sz2, HSER/HSDR codes, or static vs dynamic alloc
Use when integrating, porting, or debugging micro-ecc (uECC) ECDH, ECDSA, key generation, uECC_set_rng, signatures, curve selection, or key/signature byte formats on an MCU
Use when debugging OpenOCD, J-Link, ST-Link, CMSIS-DAP, probe connection, reset scripts, flash algorithms, GDB attach, semihosting, RTT, or SWD/JTAG failures
Use when integrating, refactoring, or debugging PLOOC object-oriented C - def_class encapsulation, private_member protection, vtable dispatch, or inheritance-style structs
Use when integrating or debugging Li-ion charger/fuel gauge ICs (BQ24295, BQ27441, MAX17048) over I2C, covering charge start, watchdog host mode, NTC/JEITA faults, and SOC jumps
| name | skill-writing-guide |
| description | Use when creating, reviewing, or refining Agent Skills for this repository |
Use this guide to create Agent Skills that are easy to discover, quick to load, and reliable in repeated use. A skill should teach an agent how to perform a task, not record how one person solved it once.
Use this skill when:
skills/<name>/SKILL.md.Do not use it for one-off project notes, private runbooks, or generic documentation that agents can already infer from standard tools.
Define the trigger.
Write the frontmatter description as a Use when... sentence that describes the situation where the skill should load.
Keep the skill operational. Prefer short instructions, decision points, verification steps, and concrete examples. Avoid essays and background stories.
Separate heavy material.
Keep SKILL.md focused. Move long API references, generated examples, or reusable scripts into supporting files beside the skill.
Add one strong example. Use a realistic example that demonstrates the workflow end-to-end. Do not add several shallow examples.
Add verification. Explain how the agent should prove the skill was used correctly before claiming completion.
Run repository validation.
Run python3 scripts/validate-skills.py before opening a pull request.
SKILL.md starts with YAML frontmatter.name and description.name matches the folder name.Use when.Before publishing a new or updated skill:
python3 scripts/validate-skills.py and confirm zero errors.name matches the folder name exactly.description starts with Use when and describes the trigger situation.Bad description:
description: Explains how to gather context, write a document, run checks, and open a PR
Better description:
description: Use when drafting release notes from repository commits and pull requests
The better version tells the agent when to load the skill without tempting it to skip the actual instructions.