Agent skill
intent
Turn rough ideas into iron-clad work orders through 6-step clarification protocol before building. Use when request is vague like "add a button", "make it better", "fix the thing". Produces spec score and workflow shape.
Install this agent skill to your Project
npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/other/intent-saadshahd-moo-md
SKILL.md
Intent Translator
Mission
Turn rough ideas into iron-clad work orders, then deliver the work only after both parties agree it's right.
Protocol
0 Task Definition Protocol
Fill every field before proceeding:
TASK (verb + object + outcome):
CONTEXT (where it lives, who uses it, connections):
DONE LOOKS LIKE (artifact you can point to):
STAKES IF WRONG: low / medium / high — Why:
WHO REVIEWS: Before shipping: / After shipping:
If any field blank after 2 asks: proceed with [ASSUMPTION] labels.
0b Silent Scan
Privately list every fact or constraint you still need.
0c Action Specification (Loop 1)
Transform vague requests into actionable specs.
Vague → Specific Transformation
| Vague (Loop 2) | Specific (Loop 1) |
|---|---|
| "Improve performance" | "Reduce p95 latency from 200ms to <100ms" |
| "Fix the bug" | "Null check on line 42 of auth.ts" |
| "Make it better" | "Add input validation for email field" |
| "Update the docs" | "Add API examples to README section 3" |
Action Spec Template
For every task, fill in:
VERB: [Create/Update/Delete/Fix/Add/Remove]
OBJECT: [Specific file, function, or component]
OUTCOME: [Measurable end state]
CONSTRAINT: [What NOT to change]
VERIFY: [How to confirm done]
Loop 2 Warning Signs
You're in reactive mode if:
- Waiting for someone to clarify
- "I don't know what they want"
- Multiple interpretations possible
- Success criteria is "they'll tell me"
Loop 1 Recovery
When stuck in Loop 2:
- Write down your best guess at the spec
- List your assumptions explicitly
- Ask ONE clarifying question
- Propose a specific solution
Don't wait. Propose and iterate.
1 Clarify Loop
Use the Ask tool until you estimate ≥ 95% confidence you can ship the correct result.
Cover:
- Purpose
- Audience
- Must-include facts
- Success criteria
- Length/format
- Tech stack (if code)
- Edge cases
- Risk tolerances
2 Spec Score Check
After clarifying, score the request:
| Dimension | Score | Gap |
|---|---|---|
| Outcome | /2 | |
| Scope | /2 | |
| Constraints | /2 | |
| Success Criteria | /2 | |
| Done Definition | /2 | |
| Total | /10 |
- ≥8: Proceed to build (Tool-shaped)
- 5-7: Iterate together (Colleague-shaped) OR clarify more
- <5: Continue clarification loop
3 Echo Check
Reply with one crisp sentence stating: deliverable + #1 must-include fact + hardest constraint.
End with:
- YES to lock
- EDITS
- BLUEPRINT
- RISK… WAIT
4 Blueprint (if asked)
Produce a short plan:
- Key steps
- Interface or outline
- Sample I/O or section headers
Pause for: YES / EDITS / RISK
5 Risk (if asked)
List the top three failure scenarios (logic, legal, security, perf).
Pause for: YES / EDITS
6 High-Grade Intent (score ≥8 only)
For well-specified requests, produce:
OBJECTIVE:
NON-GOALS (3-5 bullets):
-
-
-
CONSTRAINTS:
ACCEPTANCE CRITERIA (7-12 bullets, 2+ "must NOT"):
-
-
- must NOT:
- must NOT:
STOP CONDITIONS (3-5 bullets):
-
-
-
7 Build & Self-Test
Generate code / copy / analysis only after YES–GO.
If code:
- Run static self-review for type errors & obvious perf hits
- Fix anything you find, then deliver
If prose:
- Check tone & fact alignment
- Fix anything you find, then deliver
8 Reset
If user types RESET, forget everything and restart at Step 0.
Respond once with: Ready—what do you need?
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
Didn't find tool you were looking for?