Update to ACA

{
  "description": "Create controlled ACA copy variants for outbound experiments. Use when the user asks for spintax, subject line variants, DM variations, A/B tests, or alternate angles for a campaign. Writes single-variable variants in chat, ready to paste into the ACA campaign or sequence builder.",
  "included_files": [],
  "name": "aca-copy-variants",
  "skill_md_contents": "---\nname: aca-copy-variants\ndescription: \"Create controlled ACA copy variants for outbound experiments. Use when the user asks for spintax, subject line variants, DM variations, A/B tests, or alternate angles for a campaign. Writes single-variable variants in chat, ready to paste into the ACA campaign or sequence builder.\"\nlicense: MIT\ncompatibility: Requires the ACA connector (OAuth sign-in, no API key).\nmetadata:\n  author: ACA\n  version: \"1.1\"\n  homepage: https://www.automatedclientacquisition.com/mcp\n---\n\n# ACA copy variants\n\nCreate variations that isolate one variable at a time, so the result tells the user something. The work happens in chat; the user applies approved variants in the ACA app.\n\n## Rules\n\n- Change one thing per experiment: pain, proof, CTA, subject, or opening hook.\n- Do not create many random variants. 2 to 4 is enough.\n- For LinkedIn campaigns, write variants as message variations or prompt variations for the builder's steps.\n- For email sequences, use an A/B test step only for a real single-variable split. Subject and body variations go in the same step.\n- Spintax: keep each option a natural alternative of the same meaning, no more than 2 to 3 options per slot, never inside merge variables.\n- The connector cannot edit campaigns or sequences. Never say variants were applied.\n- Never invent past results. If the user wants to base a test on performance, use the counts ACA returns.\n\n## Workflow\n\n### 1. Identify the target\n\nIf the user names an ACA asset, find it with `list_campaigns` or `list_sequences`, then confirm with `get_campaign` or `get_sequence` (name, status, step count). The connector does not return step copy; ask the user to paste the control message.\n\nFor a LinkedIn campaign that has been running, `get_campaign_metrics` gives counts (contacted, replied, connection rate) to anchor the hypothesis. Report them as returned.\n\n### 2. Set the hypothesis and metric\n\nOne sentence: \"Changing {variable} from {control} to {variant} will raise {metric} because {reason}.\" Pick one metric: open rate (subject tests only), reply rate, positive reply rate, or connection acceptance.\n\nRough sample guidance: each variant needs enough sends to tell apart (a few hundred per arm for reply rate). Say so if the list is too small, using the list's `lead_count` from `get_lead_list` when available.\n\n### 3. Write 2 to 4 variants\n\n- A: Control (unchanged)\n- B: Alternate pain, proof, CTA, subject, or hook (only the chosen variable changes)\n- C: Optional second alternative of the same variable\n\nKeep length, tone, and everything else identical across variants.\n\n### 4. QA\n\nRun the variants through `aca-copy-spam-checker`.\n\n### 5. Hand off\n\n- LinkedIn: add the variants to the step in `https://www.automatedclientacquisition.com/campaigns/{campaign_id}`. Campaigns should be draft or paused before editing.\n- Email: add an A/B test step or subject/body variations at `https://www.automatedclientacquisition.com/email/sequences`.\n- Experiment note: the user can save the hypothesis, metric, and stop rule at `https://www.automatedclientacquisition.com/assets?tab=strategies`.\n\n## Output format\n\n```text\nVariant plan: {campaign_or_sequence}\nHypothesis: {hypothesis}\nVariable: {pain|proof|CTA|subject|hook}\nMetric: {metric}\nSample needed: {per arm}\n\nA (control): {copy}\nB: {copy}\nC: {optional}\n\nApply in: {app link}\n```\n\n## Skill chaining\n\nPreserve the workspace, relevant IDs, the user's brief, approval state, and the variants when continuing into another ACA skill.\n\n**Upstream**\n- Called by `aca-campaign-copywriting`, `aca-experiment-design`, or `aca-weekly-rhythm`.\n\n**Auto-continue conditions**\n- After variants are drafted: continue to `aca-copy-spam-checker`.\n- The experiment needs a tracking plan: continue to `aca-experiment-design`.\n\n**Stop before chaining when**\n- The user has not approved the variants.\n- The next step is applying them in the ACA app.\n\n**Downstream skills**\n- `aca-copy-spam-checker`: validate variants.\n- `aca-experiment-design`: define the single-variable test.\n- `aca-launch-outreach`: use approved variants in a new campaign.\n\n**Handoff block**\n\n```text\nChain state: {continue|needs_approval|blocked|complete}\nNext skill: {aca-skill-name|none}\nReason: {why}\nCarry forward: {workspace, campaign_id, sequence_id, hypothesis, variants, approvals, constraints}\n```\n\n## ACA tools used\n\n- `list_campaigns`, `get_campaign`, `get_campaign_metrics`\n- `list_sequences`, `get_sequence`\n- `get_lead_list`\n\n## ACA app pages\n\n- LinkedIn variants: `https://www.automatedclientacquisition.com/campaigns/{campaign_id}`\n- Email variants and A/B steps: `https://www.automatedclientacquisition.com/email/sequences`\n- Experiment note: `https://www.automatedclientacquisition.com/assets?tab=strategies`\n"
}