intent-compiler
/intent-compiler
Silently turns the owner's plain, casual message into an effective internal brief (goal, context, acceptance, constraints, mode, skills, output) before acting - the owner never writes or sees a prompt. Use on every owner message that asks for work, feedback or a decision; skip pure approvals ("pode", "sim") and slash commands.
- Author
- lucasouzadev ↗
- Repository
- agent-kit-skill ↗
- Licence
- MIT
- Last repository change
- 12.11.2026
Assessment and security
- Files in this package
- 2
- Scripts
- 0
- HTML comments
- 0
- URLs in SKILL.md
- 0
No signals detected. Repository changed on 12.11.2026, without references to retired models or APIs.
MIT, open: use and modify while retaining author attribution.
What it does and when to use it
Silently turns the owner's plain, casual message into an effective internal brief (goal, context, acceptance, constraints, mode, skills, output) before acting - the owner never writes or sees a prompt. Use on every owner message that asks for work, feedback or a decision; skip pure approvals ("pode", "sim") and slash commands.
Triggers from the description
Original author description
Silently turns the owner's plain, casual message into an effective internal brief (goal, context, acceptance, constraints, mode, skills, output) before acting - the owner never writes or sees a prompt. Use on every owner message that asks for work, feedback or a decision; skip pure approvals ("pode", "sim") and slash commands.
How it works
- Compile (in your head or scratch, never in the reply)
- Then act on the brief
- Never
SKILL.md preview
---
name: intent-compiler
description: Silently turns the owner's plain, casual message into an effective internal brief (goal, context, acceptance, constraints, mode, skills, output) before acting - the owner never writes or sees a prompt. Use on every owner message that asks for work, feedback or a decision; skip pure approvals ("pode", "sim") and slash commands.
---
# Intent compiler (invisible)
The owner talks like in a meeting. You turn that talk into the prompt a prompt engineer would have written — **for yourself**, before acting. Rubric and examples: [rubric.md](rubric.md).
## Compile (in your head or scratch, never in the reply)
```
GOAL verb + observable result, one line
CONTEXT what the owner said + what the repo says: handoff, last commits, files/screens named or implied
(search to resolve "aquilo", "a tela nova", "o bug de ontem"); quote the owner's words for intent
ACCEPTANCE how we both know it is done, in numbers or visible behaviour
CONSTRAINTS AGENTS.md rules that apply (approval gate, build/release rules, design system, security)
OUT OF SCOPE what not to touch, even if tempting
MODE Fixer | Builder | Product | Architect | Lens | Auditor | Status
SKILLS the 1–3 skills/docs to load (e.g. systematic-debug, ux-check, security-core); media attached → media-investigation; 3+ asks or a list → triage
UNKNOWNS only what changes the result and the repo cannot answer → at most ONE question
OUTPUT what the owner gets back: plan for approval | answer | status | evidence
```
## Then act on the brief
- Enough to act → do the next step (the approval gate still applies before any write).
- One blocking unknown → ask it as a multiple-choice question, recommendation first. Fill everything else with stated defaults ("I assumed X; say if not", in the owner's language).
- Restate only GOAL + ACCEPTANCE to the owner, in one line, when the task is non-trivial. The rest of the brief stays invisible.
- The brief is not stored in files; GOAL + ACCEPTANCE + approvals go to the handoff when the session closes.
## Never
Show the brief unless the owner asks ("mostra o brief", "como você entendeu?"). Ask the owner to rephrase as a prompt. Invent context the repo does not confirm — mark it as an assumption.
Files in this package (2)
- SKILL.md 2 KB
- rubric.md 2 KB
Versions
The database has no other skills named intent-compiler.
Database snapshot: 13.11.2026 · package version 314efb68b8b6 (SHA-256 digest of the entire package)
Downloads in the last 30 days: 0 · ranking
Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.
Similar skills
- 90sharpen-prompt
Rewrite a vague request into a precise prompt — goal, context, scope, constraints, what done looks like, output format. Use when the user asks to improve, rewrite, sharpen or check a prompt. - 100user-intent-priority
User intent recognition and response priority standard. Requires classifying every user message before acting, answering questions before executing, pausing autonomous work when the user speaks, and gating tool calls on actual authorization.