convert-agent-skill-to-holistics

Étoiles7

Forks0

Mis à jour29 juillet 2026 à 12:02

Convert an agent skill written for another runtime (Claude `SKILL.md`, OpenAI/Codex skill, raw system prompt) into a Holistics AML `Skill` block — or the reverse direction. Use whenever the user wants to migrate, port, or translate a skill across formats, including phrasings like "convert this Claude skill to AML", "port this OpenAI skill to Holistics", "turn this SKILL.md into an AML Skill", "translate this AML skill back to Claude", or "migrate this agent skill".

Installation

Installer avec Codex ou Claude Copiez ce prompt, collez-le dans Codex, Claude ou un autre assistant, puis laissez-le vérifier la page du skill et l'installer pour vous.

Source

holistics

holistics/skills

Ouvrir le dépôt GitHub Voir les dépôts du créateur

Téléchargement

Métiers associésSOC

Basé sur la classification professionnelle SOC

Autres occupations informatiquesProfessions informatiques et mathématiques·SOC 15-1299

Explorateur de fichiers

2 fichiers

SKILL.md

readonly

name

convert-agent-skill-to-holistics

description

Converting an agent skill into a Holistics AML Skill

Use this skill when porting an agent skill between Holistics AML and another runtime (Claude SKILL.md, OpenAI/Codex skill, raw system prompt). What changes is the wrapper, the file path, and a handful of field semantics; the skill's purpose and content stay the same.

For the principles that govern how a good Holistics skill is written — description triggers, concise body, explain-the-why — see the create-holistics-skill skill. Apply those after the mechanical conversion below.

Features that map directly

Holistics AML	Claude / OpenAI Markdown `SKILL.md`
`Skill <name> { ... }` block	YAML frontmatter + Markdown body
`description: "..."`	frontmatter `description:`
`content: @md ... ;;`	Markdown body below the frontmatter
`invocation` field	see invocation tables below
`disabled: true`	delete the file (or the equivalent field if the host runtime supports it)
Placed at `settings/ai/skills/<name>/skill.aml`	Placed at the consuming tool's project-skill path (commonly `.claude/skills/<name>/SKILL.md`)

Invocation mapping — from Claude

`disable-model-invocation`	`user-invocable`	AML Skill
false (default)	false	`invocation: 'auto'`
false (default)	true (default)	`invocation: 'auto'`
true	true (default)	`invocation: 'manual'`
true	false	`disabled: true, invocation: 'manual'`

Invocation mapping — from OpenAI/Codex

`allow_implicit_invocation`	AML Skill
true (default)	`invocation: 'auto'`
false	`invocation: 'manual'`

Handling incompatibilities

Agent skills and Holistics skills cover different sets of functionalities. For example:

Agent skills can bundle references/* documents and scripts/* executables; Holistics skills currently cannot.
Holistics skills can be parameterized via the programmatic AML Slot[...] mechanism; agent skills cannot.

The exact list of differences may change over time; treat any feature in the source skill not covered by the mapping tables above as an incompatibility.

When the source skill uses an incompatible feature, do not silently drop, inline, or translate it. The right move depends on how that feature is being used in the source skill, and only the user can tell you.

For each incompatible feature found:

State which feature it is and why it has no equivalent in the target.
Quote or describe what the feature is doing in the source skill (the Slot's purpose, the script's behavior, the role-gate's condition).
Ask the user how to handle it in the target. Don't suggest a specific answer; let them decide.
Apply the user's answer and continue. If multiple incompatibilities exist, ask about each separately — they may want different treatment.

If the user opts to defer, emit the converted skill without the incompatible feature and call out in the conversion summary exactly what was left out, so they can rebuild it elsewhere (MCP tool, dataset, separate skill) later.

Conversion workflow

Capture the source skill's name and description.
Rewrite the description for Holistics AI's triggering surface — concrete phrasings, what + when. Don't paste the original verbatim; trigger surfaces differ between agents.
Move the body content into content: @md ... ;;. Strip any "when to use" lines that have migrated into the body (in Holistics, those belong in description).
Map invocation fields using the tables above.
Scan the source skill for any feature that isn't covered by the mapping tables above. For each one found, halt and ask the user — see Handling incompatibilities. Resume the workflow with their decisions applied.
Place at settings/ai/skills/<name>/skill.aml (or the target tool's project-skill path if going the other direction).

Plus depuis ce dépôt

même dépôt

review-chat

holistics/skills

Review a past conversation between a Holistics data-analyst AI agent and a user, then produce insight report with three sections (`Summary`, `Recap`, and `Suggestions`). Use this whenever the user wants to assess, grade, debrief, or critique how the agent performed on a chat, including phrasings like " this conversation", "how did the agent do", "grade this chat", "what could the AI have done better", or "what context should I add to improve the agent", even if they don't say the word "review".

2026-07-297

apply-chat-suggestions

holistics/skills

Address suggestions in a past conversation's insight by performing each suggestion's actionable fix. Use when the user asks to implement, apply, address, or act on the suggestions from a conversation insight. Requires a conversation ID. Do NOT use this to generate a new insight.

2026-07-297

build-dashboard-theme

holistics/skills

Create, update, or apply a Holistics dashboard theme (library/themes/) from any input: a brand name, website, screenshot, brand guidelines, explicit colors, or a mood description. Use when the user wants to theme, brand, restyle, or change a dashboard's look and feel. Typical phrasings: build a theme, match our brand, use these colors, make it dark, make it warm/minimal/premium/editorial, glow up this dashboard. Do NOT trigger for building a dashboard, authoring block content, or adding controls.

2026-07-297

create-holistics-skill

holistics/skills

Author a Holistics skill. Use whenever the user wants to write, edit, or refine a skill that gives Holistics AI specific knowledge, workflows, or guardrails — including phrasings like "add a skill for X", "create a workspace glossary skill", "turn this prompt into a reusable skill", or "make this Holistics AI smarter about Y". For converting a skill written for another runtime (Claude, OpenAI/Codex) into AML, use `convert-agent-skill-to-holistics` instead.

2026-07-297

build-custom-chart

holistics/skills

Create or refine a Holistics custom chart (CustomChartDef in library/custom_charts/) for a chart type the built-in viz can't cover. Use when the user wants to create a custom chart, convert a Vega/Vega-Lite spec into Holistics AML, or scaffold or refine a chart template — or when a dashboard needs a chart form with no built-in equivalent (heatmap matrix, sankey, waterfall, box plot, ...). Typical phrasings: create a custom chart for X, turn this Vega-Lite spec into a Holistics chart, we need a sankey of X. Do NOT trigger for a standard built-in chart or a dynamic content block.

2026-07-297

build-dashboard

holistics/skills

Build or edit a Holistics canvas dashboard (`page.aml` / a `Dashboard {}` object) — its blocks, layout, filter interactions, and theme. Use when building a dashboard or report page, adding or arranging blocks, or adding dashboard-level filters. Typical phrasings: build a dashboard for X, make a dashboard showing X, add a chart or KPI or table to a dashboard, add a filter or date control, add or edit a title or section header. Do NOT trigger for answering a data question on its own — a one-off query, chart, or analysis with no dashboard to put it on — or for editing models, datasets, or metric definitions, authoring a reusable custom chart type or theme object, or restyling a dashboard's look.

2026-07-297

name

convert-agent-skill-to-holistics

description

Converting an agent skill into a Holistics AML Skill

Features that map directly

Holistics AML	Claude / OpenAI Markdown `SKILL.md`
`Skill <name> { ... }` block	YAML frontmatter + Markdown body
`description: "..."`	frontmatter `description:`
`content: @md ... ;;`	Markdown body below the frontmatter
`invocation` field	see invocation tables below
`disabled: true`	delete the file (or the equivalent field if the host runtime supports it)
Placed at `settings/ai/skills/<name>/skill.aml`	Placed at the consuming tool's project-skill path (commonly `.claude/skills/<name>/SKILL.md`)

Invocation mapping — from Claude

`disable-model-invocation`	`user-invocable`	AML Skill
false (default)	false	`invocation: 'auto'`
false (default)	true (default)	`invocation: 'auto'`
true	true (default)	`invocation: 'manual'`
true	false	`disabled: true, invocation: 'manual'`

Invocation mapping — from OpenAI/Codex

`allow_implicit_invocation`	AML Skill
true (default)	`invocation: 'auto'`
false	`invocation: 'manual'`

Handling incompatibilities

Agent skills and Holistics skills cover different sets of functionalities. For example:

Agent skills can bundle references/* documents and scripts/* executables; Holistics skills currently cannot.
Holistics skills can be parameterized via the programmatic AML Slot[...] mechanism; agent skills cannot.

The exact list of differences may change over time; treat any feature in the source skill not covered by the mapping tables above as an incompatibility.

For each incompatible feature found:

State which feature it is and why it has no equivalent in the target.
Quote or describe what the feature is doing in the source skill (the Slot's purpose, the script's behavior, the role-gate's condition).
Ask the user how to handle it in the target. Don't suggest a specific answer; let them decide.
Apply the user's answer and continue. If multiple incompatibilities exist, ask about each separately — they may want different treatment.

Conversion workflow

Capture the source skill's name and description.
Rewrite the description for Holistics AI's triggering surface — concrete phrasings, what + when. Don't paste the original verbatim; trigger surfaces differ between agents.
Move the body content into content: @md ... ;;. Strip any "when to use" lines that have migrated into the body (in Holistics, those belong in description).
Map invocation fields using the tables above.
Scan the source skill for any feature that isn't covered by the mapping tables above. For each one found, halt and ask the user — see Handling incompatibilities. Resume the workflow with their decisions applied.
Place at settings/ai/skills/<name>/skill.aml (or the target tool's project-skill path if going the other direction).