to-prompt — Claude Code skill: a brief for another LLM
Turns a task into a written brief under docs/prompts/, so a different model can carry out the implementation.
npx skills add pedronauck/skills --skill to-prompt -a claude-codeWhat to-prompt does
When one agent hands work to another, the context usually gets lost. This skill writes it down instead. The result is a markdown file at docs/prompts/<YYYYMMDD-HHmm>_<slug>.md covering the problem, the current state, the requirements and the constraints. It does not pick a solution — that call belongs to whoever receives the brief.
The procedure runs in stages. First you classify the work as a bug fix, an improvement or a feature. Then comes evidence gathering: read the code paths involved, quote them as they stand, capture the relevant log and error lines while leaving credentials out. The skeleton comes from assets/brief-template.md. A final pass checks that decisions the user already approved survived intact and that no new mandates crept in.
The skill states its own limits: no one-shot questions, no end-user copy. Its frontmatter also sets disable-model-invocation: true, which means the agent will not trigger it on its own — you start it.
What it helps with
- Hand a task to another agent with enough context
- Lock in approved decisions before coding starts
- Collect code excerpts and logs into one file
- Keep requirements apart from implementation suggestions
- Save the brief with a date and readable name in the repo
Example prompts for the agent
- Write a brief for the auth redirect bug and save it under
docs/prompts/ - Turn this task into a handoff brief for another agent, use
assets/brief-template.md - Draft a brief for the new export feature, keep the approved decisions as they are
How to install to-prompt
1Skills CLI (Vercel)
npx skills add pedronauck/skills --skill to-prompt -a claude-code2Manually: put the files in place
~/.claude/skills/to-prompt/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/to-prompt/
1Skills CLI (Vercel)
npx skills add pedronauck/skills --skill to-prompt -a codex2Manually: put the files in place
.agents/skills/to-prompt/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add pedronauck/skills --skill to-prompt -a cursor2Manually: put the files in place
.cursor/skills/to-prompt/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install pedronauck/skills to-prompt2Manually: put the files in place
.github/skills/to-prompt/SKILL.md1Skills CLI (Vercel)
npx skills add pedronauck/skills --skill to-prompt -a gemini-cli2Platform CLI
gemini skills install https://github.com/pedronauck/skills --path skills/mine/to-promptManually: put the files in place
.devin/skills/to-prompt/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add pedronauck/skills --skill to-prompt -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about to-prompt
How do I install to-prompt in Claude Code?
Run npx skills add pedronauck/skills --skill to-prompt -a claude-code. The skill file lands at ~/.claude/skills/to-prompt/SKILL.md.
What does to-prompt do?
It writes an implementation brief and stores it in docs/prompts/. The brief holds the problem, current state, requirements and constraints, while the receiving model chooses the approach.
Which tasks does the skill leave out?
One-shot questions and copy meant for end users. Neither needs a context-rich implementation brief.
