# Send Taste to your agent
Hand the extracted package to your coding agent with a concrete install brief instead of figuring it out manually.
## Fast path
- Download the package from Yavira.
- Extract it into a folder your agent can access.
- Paste one of the prompts below and point your agent at the extracted folder.
## Suggested prompts
### New install

```text
I downloaded a skill package from Yavira. Read SKILL.md from the extracted folder and install it by following the included instructions. Tell me what you changed and call out any manual steps you could not complete.
```
### Upgrade existing

```text
I downloaded an updated skill package from Yavira. Read SKILL.md from the extracted folder, compare it with my current installation, and upgrade it while preserving any custom configuration unless the package docs explicitly say otherwise. Summarize what changed and any follow-up checks I should run.
```
## Machine-readable fields
```json
{
  "schemaVersion": "1.0",
  "item": {
    "slug": "taste",
    "name": "Taste",
    "source": "tencent",
    "type": "skill",
    "category": "AI 智能",
    "sourceUrl": "https://clawhub.ai/ivangdavila/taste",
    "canonicalUrl": "https://clawhub.ai/ivangdavila/taste",
    "targetPlatform": "OpenClaw"
  },
  "install": {
    "downloadUrl": "/downloads/taste",
    "sourceDownloadUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=taste",
    "sourcePlatform": "tencent",
    "targetPlatform": "OpenClaw",
    "packageFormat": "ZIP package",
    "primaryDoc": "SKILL.md",
    "includedAssets": [
      "SKILL.md",
      "antipatterns.md",
      "development.md",
      "learning.md",
      "prompting.md",
      "visual.md"
    ],
    "downloadMode": "redirect",
    "sourceHealth": {
      "source": "tencent",
      "slug": "taste",
      "status": "healthy",
      "reason": "direct_download_ok",
      "recommendedAction": "download",
      "checkedAt": "2026-05-04T01:23:36.382Z",
      "expiresAt": "2026-05-11T01:23:36.382Z",
      "httpStatus": 200,
      "finalUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=taste",
      "contentType": "application/zip",
      "probeMethod": "head",
      "details": {
        "probeUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=taste",
        "contentDisposition": "attachment; filename=\"taste-1.0.0.zip\"",
        "redirectLocation": null,
        "bodySnippet": null,
        "slug": "taste"
      },
      "scope": "item",
      "summary": "Item download looks usable.",
      "detail": "Yavira can redirect you to the upstream package for this item.",
      "primaryActionLabel": "Download for OpenClaw",
      "primaryActionHref": "/downloads/taste"
    },
    "validation": {
      "installChecklist": [
        "Use the Yavira download entry.",
        "Review SKILL.md after the package is downloaded.",
        "Confirm the extracted package contains the expected setup assets."
      ],
      "postInstallChecks": [
        "Confirm the extracted package includes the expected docs or setup files.",
        "Validate the skill or prompts are available in your target agent workspace.",
        "Capture any manual follow-up steps the agent could not complete."
      ]
    }
  },
  "links": {
    "detailUrl": "https://openagent3.xyz/skills/taste",
    "downloadUrl": "https://openagent3.xyz/downloads/taste",
    "agentUrl": "https://openagent3.xyz/skills/taste/agent",
    "manifestUrl": "https://openagent3.xyz/skills/taste/agent.json",
    "briefUrl": "https://openagent3.xyz/skills/taste/agent.md"
  }
}
```
## Documentation

### Core Principle — Learn From the Human

You don't have taste yet. The human does. Your job is to:

Ask when you don't understand why something is good/bad
Record every correction and explanation
Apply learned patterns to future judgments
Check your judgments against the human's until calibrated

Taste is learned through exposure + feedback. The human provides both.

### Workspace

Store taste learning in ~/taste/:

corrections/ — Each time human corrects your judgment
preferences/ — Human's stated aesthetic preferences by domain
patterns/ — Extracted rules from accumulated corrections
calibration.md — Current confidence level per domain

### The Learning Loop

When evaluating anything aesthetic:

State your judgment — "I think X because Y"
Ask for feedback — "Does this match your taste? What am I missing?"
If corrected:

Ask WHY (genuinely curious, not defensive)
Record the correction with context
Extract the underlying pattern
Update your calibration confidence

Never defend your aesthetic judgment against the human's. Learn from the gap.

### Genuine Curiosity Protocol

When the human says something is better/worse than you thought:

Ask specifically:

"What makes this work better than the alternative?"
"What am I not seeing here?"
"Is this a general principle or specific to this context?"
"Would this apply to [similar situation]?"

Don't ask vaguely:

❌ "Can you explain more?"
❌ "Why do you think that?"

Specific questions show you're trying to extract transferable knowledge.

### Recording Corrections

When human corrects your taste judgment:

Date: [timestamp]
Domain: [design/writing/etc]
My judgment: [what I said]
Human's correction: [what they said]
Why (their explanation): [the reasoning]
Pattern extracted: [generalizable rule]
Confidence update: [how this changes my calibration]

Store in corrections/[domain]/[date].md

### Calibration Levels

Track your confidence per domain:

LevelMeaningBehaviorUncalibratedNo feedback yetAlways ask, never assertLearningSome corrections receivedState tentatively, ask for confirmationCalibratingPatterns emergingState with reasoning, check occasionallyCalibratedConsistent agreementState confidently, still open to correction

Start uncalibrated in every domain. Earn confidence through accurate predictions.

### Load Reference When Needed

SituationReferenceFull learning system and calibration processlearning.mdEvaluating visual/design workvisual.mdEvaluating writing/prosewriting.mdUnderstanding taste development theorydevelopment.mdRecognizing bad taste patternsantipatterns.mdGenerating tasteful creative outputprompting.md

These are starting points. Human feedback overrides everything in them.
## Trust
- Source: tencent
- Verification: Indexed source record
- Publisher: ivangdavila
- Version: 1.0.0
## Source health
- Status: healthy
- Item download looks usable.
- Yavira can redirect you to the upstream package for this item.
- Health scope: item
- Reason: direct_download_ok
- Checked at: 2026-05-04T01:23:36.382Z
- Expires at: 2026-05-11T01:23:36.382Z
- Recommended action: Download for OpenClaw
## Links
- [Detail page](https://openagent3.xyz/skills/taste)
- [Send to Agent page](https://openagent3.xyz/skills/taste/agent)
- [JSON manifest](https://openagent3.xyz/skills/taste/agent.json)
- [Markdown brief](https://openagent3.xyz/skills/taste/agent.md)
- [Download page](https://openagent3.xyz/downloads/taste)