spyderweb47 My Template Skill
Purpose
Replace this paragraph with one or two sentences describing what this skill does and the user-facing outcome it produces. Keep it concrete — "generates an equity curve from fundamental data" is better than "does financial stuff".
When to use this skill
Bullet list of situations where Vibe Trade should dispatch to this skill:
- When the user asks for ...
- When context contains ...
- When the chart panel provides ...
Be specific. If two skills could plausibly handle the same message, the "when to use" sections decide which one Vibe Trade picks.
Instructions
Numbered, imperative steps that describe exactly what the skill's processor should do for each input shape:
- If the message is type A: do X, emit tool Y, return Z.
- If
context.foois present: branch into the Y workflow. - Otherwise: default behavior.
Keep each step testable and concrete. The less ambiguity, the less LLM-induced drift.
Inputs
| Key | Type | Meaning |
|---|---|---|
message |
string | The user's chat message |
context.<key> |
type | Whatever your skill cares about from the router |
Outputs
Describe what the SkillResponse will contain:
reply— ...script— optional ...data.*— optional ...
Tools used
| Tool | When | Payload |
|---|---|---|
script_editor.load |
When a script is generated | The JS source string |
bottom_panel.activate_tab |
After generating output | The target tab id |
notify.toast |
For transient status messages | {level, message} |
Examples
Example 1
User input: "Do the thing with the stuff."
→ What the skill returns and why.
Example 2
User input: (a more complex variant)
→ How the skill handles it differently.
Underlying implementation
This template has no Python processor yet. To wire it up:
- Add an
async def _mytemplate_processor(message, context, tools)incore/agents/processors.py - Register it in the
PROCESSORSdict:PROCESSORS = { ..., "mytemplate": _mytemplate_processor, } - Restart the backend. The
VibeTrade.dispatch("mytemplate", ...)call will now route here.
Frontmatter reference
| Field | Type | Description |
|---|---|---|
id |
string | Unique kebab-case identifier (also the folder name) |
name |
string | Full display name shown in the chip row |
tagline |
string | Short label (1 word) for compact UI |
description |
string | One-sentence description; shown as a tooltip |
version, author |
string | Informational, surfaced in a future "skill info" UI |
category |
string | analysis, generation, simulation, etc. (free-form) |
icon |
string | Icon name; resolved on the frontend. Unknown icons fall back to sparkle |
color |
string | Hex color used for the chip accent |
tools |
string[] | Allowlist of tool ids from skills/tools.py::TOOL_CATALOG |
output_tabs |
object[] | Bottom-panel tabs this skill contributes |
store_slots |
string[] | Documentation-only: store keys this skill writes |
input_hints.placeholder |
string | Chat textarea placeholder when this is the only active skill |
input_hints.supports_fingerprint |
bool | true if your skill accepts SHAPE fingerprints from the chart selector |
Available tools
See skills/tools.py::TOOL_CATALOG for the full list. The tool categories
are:
chart.drawing.*— activate a drawing tool (trendline, rectangle, fib, etc.)chart.pattern_selector— open the pattern selector to capture a fingerprintchart.highlight_matches/chart.draw_markers— chart overlayschart.focus_range/chart.set_timeframe— chart navigationchatbox.card.strategy_builder— inject the strategy builder form in chatchatbox.card.generic— inject a generic card with title/body/actionsbottom_panel.activate_tab/bottom_panel.set_data— bottom panel controlscript_editor.load/script_editor.run— code editor controldata.indicators.add/data.indicators.toggle— indicator registrynotify.toast— user notifications
Only declare tools in the frontmatter that your processor actually uses. The frontend tool registry rejects any tool_call whose id isn't in the skill's declared list.