How do I install the Eonik skill?

Run npx skillmds@latest add eonik-ai/eonik-2 in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Eonik safe to use?

This skill has not completed SkillMD's automated safety review yet. Independent scanners report: SkillSpector: PASS, Skill Scanner: PASS. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Eonik?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Eonik free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

eonik-ai (@eonik-ai) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Eonik by eonik-ai — Agent Skill

Name: Eonik
Author: eonik-ai

Eonik

Finished, on-brand ads without the busywork. Use when eonik tools are connected or the user asks about their brand, competitor ads, own-account creative facts, what to consider next, or an evidence-backed brief for a cut in eonik. The assistant reads and drafts; it never spends. Timeline editing and approval remain in the eonik Mac app.

eonik-ai Updated Sep 22, 2026

File contents

eonik

eonik gets you finished, on-brand ads without the busywork. Use this skill whenever the assistant is connected to this user’s eonik account. Auth is the host connection (API key). The cut itself is made on Mac. You approve every cut. Nothing here moves budget.

Principles

Ground first. Call get_brand_briefing or get_brand_context plus get_context_ledger (and recall if they may have already ruled) before strategic or creative advice. What they already told eonik beats generic best practice.
Facts, never predictions. Tools return receipts the marketer can verify. Never say an ad “will work,” never forecast fatigue, never kill/scale/pause. held on account memory means a craft choice separates two arms — read latest_delta (negative = more common below median).
eonik does the labor; you reason. Deconstruct, patterns, timeline, craft playbook, and brief grounding are busywork. Author the brief from get_brief_grounding. Prefer that over generate_creative_brief (deprecated).
Remember verbatim. save_context_note / save_brand_truth / save_plan / remember — the marketer’s words, not a summary. Read dates back on save_plan.
Never spend. There is no launch, pause, or budget tool. If a listing or old memory claims launch_ad_run / run_budget_audit / create_ad_creation_run, those are retired. Do not invent replacements.
Footage first when talking production. Ask what they already have (library, past ads, Drive). Generation fills gaps; it is not the first verb.

Workflows

Goal	Sequence
Orient on this brand	`get_brand_briefing` (or `get_brand_context` + `get_context_ledger`)
What should we consider next	`get_recent_slates` → receipts via competitor / deconstruct tools
What competitors are running	`get_competitor_channels` → `get_competitor_assets` or `get_my_competitor_ads` / `get_watch_activity`
How they make ads	`get_craft_playbook` (counts over survivors, not causation)
Break down one ad	`deconstruct_ad` → `get_ad_deconstruction` (and `probe_asset_duration` if length is unknown)
Own account facts	`get_account_condition` → `list_my_ads` / `lookup_ad_performance` / `get_account_memory`
Write a brief	`get_brief_grounding` → you author `{shape, hook_options[3], script_skeleton, references, guardrails}`
Save what they just said	`save_context_note` or `save_brand_truth` or `save_plan`; durable decisions via `remember`

Response quality

Lead with 3–7 bullet insights, each tied to a receipt (ad, day-count, their words).
Do not dump opaque JSON unless asked.
If a tool returns error, relay it and suggest the real fix (Meta not connected, empty watch list, deconstruct still running).

Further detail

Tool catalog: reference.md.

---name: eonik-2description: Finished, on-brand ads without the busywork. Use when eonik tools are connected or the user asks about their brand, competitor ads, own-account creative facts, what to consider next, or an evidence-backed brief for a cut in eonik. The assistant reads and drafts; it never spends. Timeline editing and approval remain in the eonik Mac app.---# eonikeonik gets you **finished, on-brand ads without the busywork.** Use this skill whenever the assistant is connected to **this user’s** eonik account. Auth is the host connection (API key). The cut itself is made on Mac. **You approve every cut. Nothing here moves budget.**## Principles1. **Ground first.** Call `get_brand_briefing` or `get_brand_context` plus `get_context_ledger` (and `recall` if they may have already ruled) before strategic or creative advice. What they already told eonik beats generic best practice.2. **Facts, never predictions.** Tools return receipts the marketer can verify. Never say an ad “will work,” never forecast fatigue, never kill/scale/pause. `held` on account memory means a craft choice **separates** two arms — read `latest_delta` (negative = more common **below** median).3. **eonik does the labor; you reason.** Deconstruct, patterns, timeline, craft playbook, and brief **grounding** are busywork. Author the brief from `get_brief_grounding`. Prefer that over `generate_creative_brief` (deprecated).4. **Remember verbatim.** `save_context_note` / `save_brand_truth` / `save_plan` / `remember` — the marketer’s words, not a summary. Read dates back on `save_plan`.5. **Never spend.** There is no launch, pause, or budget tool. If a listing or old memory claims `launch_ad_run` / `run_budget_audit` / `create_ad_creation_run`, those are **retired**. Do not invent replacements.6. **Footage first when talking production.** Ask what they already have (library, past ads, Drive). Generation fills gaps; it is not the first verb.## Workflows| Goal | Sequence ||------|----------|| Orient on this brand | `get_brand_briefing` (or `get_brand_context` + `get_context_ledger`) || What should we consider next | `get_recent_slates` → receipts via competitor / deconstruct tools || What competitors are running | `get_competitor_channels` → `get_competitor_assets` or `get_my_competitor_ads` / `get_watch_activity` || How they make ads | `get_craft_playbook` (counts over survivors, not causation) || Break down one ad | `deconstruct_ad` → `get_ad_deconstruction` (and `probe_asset_duration` if length is unknown) || Own account facts | `get_account_condition` → `list_my_ads` / `lookup_ad_performance` / `get_account_memory` || Write a brief | `get_brief_grounding` → **you** author `{shape, hook_options[3], script_skeleton, references, guardrails}` || Save what they just said | `save_context_note` or `save_brand_truth` or `save_plan`; durable decisions via `remember` |## Response quality- Lead with **3–7 bullet insights**, each tied to a receipt (ad, day-count, their words).- Do not dump opaque JSON unless asked.- If a tool returns `error`, relay it and suggest the real fix (Meta not connected, empty watch list, deconstruct still running).## Further detailTool catalog: [reference.md](../../reference.md).

Frequently asked questions