Writesonic

View saved version →

---
name: action-center
description: Rank Writesonic's recommended actions by impact and pick the three with the biggest visibility lift for the least effort. Use when the user asks what to do, what to work on, recommendations, or priorities.
---
Prioritize actions for project "$ARGUMENTS" (resolve via `get_all_projects` if needed).

1. `get_actionables_dashboard` FIRST — every recommendation with per-category counts and impact. Do not call the individual category tools before this.
2. For the leading categories, pull estimated lift: `get_brand_gap_impact`, `get_new_content_impact`, `get_broken_pages_impact`.
3. Rank by lift ÷ effort (broken-page fixes are cheapest, new content is most expensive) and name the three to do first, each with the number behind it.

View saved version →

---
name: competitor-teardown
description: Full teardown of one competitor — every prompt where they're cited and we aren't, topics where they out-rank us, platforms where their lead is biggest, and a plan to overtake them. Use when the user names a competitor and asks how they beat us or how to catch up.
---
Tear down competitor "$ARGUMENTS" for the user's project (resolve the project via `get_all_projects`; confirm the competitor is tracked via `get_all_websites`).

1. `get_prompts_not_mentioning_you` — prompts where we are absent; keep the ones where this competitor appears.
2. `query_performance_report` dimensions=["website", "topic"] website_type="SELF+COMPETITORS" — topics where they out-rank us.
3. `query_performance_report` dimensions=["website", "model"] website_type="SELF+COMPETITORS" — platforms where their lead is biggest.
4. `get_brand_gap_summary` → `get_all_brand_gap_opportunities` — third-party pages citing them but not us (the outreach list).

Finish with a prioritized plan: biggest gap first, one concrete action per gap. Cite the numbers behind each claim.

View saved version →

---
name: content-gap-brief
description: Find the top content gaps by impact, optionally start tracking the underlying questions, and set up citation monitoring for the pages once published. Use when the user asks what content to create, content opportunities, or content gaps.
---
Find content gaps for project "$ARGUMENTS" (resolve via `get_all_projects` if needed).

1. `get_all_new_content_opportunities` — questions AI answers where we have no page.
2. `get_new_content_impact` — rank the gaps by estimated visibility lift; take the top three unless the user asks for more.
3. For each gap, draft a one-paragraph brief: the question, the platforms answering it, who is cited today, and the angle for our page.
4. Only if the user asks to track them: show the exact prompts you would create, ask for confirmation, then `create_prompt` for each. Never create prompts without an explicit yes.
5. Tell the user that once the pages are live, `get_webpage_citation_trends` with the page URL as the `cited_page_url` filter will show their first AI citations.

View saved version →

---
name: kpi
description: KPI snapshot for a project — visibility, mentions, citations, sentiment with deltas vs the previous period.
disable-model-invocation: true
---
Call `get_kpi_summary` for project "$ARGUMENTS" (resolve via `get_all_projects`; ask if ambiguous). Present the KPIs with their previous-period deltas as a compact table, declines first.

View saved version →

---
name: markets
description: Brand visibility broken down by geographic market.
disable-model-invocation: true
---
For project "$ARGUMENTS" (resolve via `get_all_projects` if needed) call `query_performance_report` dimensions=["market"] website_type="SELF". Show visibility_score and share_of_voice per market, highest first.

View saved version →

---
name: monthly-performance-report
description: Monthly performance summary — visibility, citation and sentiment trends for the brand. Use when the user asks for a monthly report, month-in-review, or performance summary over months.
---
Build the monthly performance summary for project "$ARGUMENTS" (resolve via `get_all_projects` if not given). Cover the last 3 full months unless the user specifies a range.

1. `query_performance_report` dimensions=["date"] date_aggregation_interval="month" website_type="SELF" — visibility_score, share_of_voice, mention_count per month.
2. `query_citation_report` dimensions=["date"] date_aggregation_interval="month" website_type="SELF" — citation count and unique pages per month.
3. `query_sentiment_report` dimensions=["date"] date_aggregation_interval="month" website_type="SELF" — sentiment per month (request `pct_positive`).

Report each layer as its own short table, then a 3-bullet takeaway. Numbers and periods only; declines first.

View saved version →

---
name: page-optimization
description: Prioritized list of changes to make a specific page easier for AI to cite, with rationale, and how to verify the lift afterwards. Use when the user gives a page URL and asks how to win more AI citations for it.
---
Optimize the page "$ARGUMENTS" (a URL; ask for one if missing) in the user's project (resolve via `get_all_projects`).

1. `get_all_optimization_opportunities` — find the page and its detected issues. If it isn't listed, say so and suggest `track_pages` first.
2. `get_optimization_fixes` for that page — concrete fixes with rationale.
3. Present the fixes in priority order: what to change, why AI cites pages that do this, expected effect.
4. Close with the verification step: after publishing, `get_webpage_citation_trends` filtered by `cited_page_url` to confirm the citation lift.

View saved version →

---
name: projects
description: List the user's Writesonic projects with their type (STANDARD or PITCH).
disable-model-invocation: true
---
Call `get_all_projects` (pass "$ARGUMENTS" as the `query` filter if given). Show a table: name, type, url, created date. Read the FULL result before answering — the list is newest-first and can be long.

View saved version →

---
name: quarterly-visibility-report
description: One-page quarterly visibility summary ready to paste into a deck — overall score and trend, top three topic gainers and losers with a one-line reason each. Use for quarterly reviews, QBRs, or "summarize the quarter".
---
Write a one-page quarterly visibility summary for project "$ARGUMENTS" (resolve via `get_all_projects` if needed). Default to the last completed quarter vs the one before; ask if ambiguous.

1. `get_kpi_summary` with the quarter as current_from/current_to and the prior quarter as previous_from/previous_to — headline numbers with deltas.
2. `query_performance_report` dimensions=["date"] date_aggregation_interval="week" website_type="SELF" — the trend line across the quarter.
3. `query_performance_report` dimensions=["topic"] website_type="SELF" with start_date/end_date = this quarter and comparison_start_date/comparison_end_date = last quarter — each topic row carries previous_* values; rank by change to get the three biggest gainers and the three biggest losers.
4. For each of those six topics, one line on why (use `get_prompts_leaderboard` filtered to the topic if the cause isn't obvious).

Output: headline → trend sentence → gainers table → losers table → one-line reasons. Keep it to one page.

View saved version →

---
name: sov
description: Share of voice vs competitors, optionally on one AI platform and over N days (e.g. "/writesonic:sov ChatGPT 30").
disable-model-invocation: true
---
Arguments "$ARGUMENTS" may contain a platform name and/or a number of days (default: all platforms, last 30 days). Resolve the project via `get_all_projects` if needed, then call `query_performance_report` with dimensions=["website"], website_type="SELF+COMPETITORS", measures=["share_of_voice","visibility_score"], the date range, and a filters.platform "is" condition if a platform was named. Rank websites by share_of_voice and state our rank and the gap to #1.

View saved version →

---
name: topics
description: Topic strength ranking — visibility by topic, strongest to weakest.
disable-model-invocation: true
---
For project "$ARGUMENTS" (resolve via `get_all_projects` if needed) call `query_performance_report` dimensions=["topic"] website_type="SELF" order_by visibility_score desc. Show the ranked table and name the strongest and weakest three.

View saved version →

---
name: visibility-drop
description: Diagnose a visibility drop — which topics and platforms declined, whether citations were lost or sentiment turned, the exact prompts responsible, and a fix per cause. Use when the user asks why visibility, share of voice, or mentions went down.
---
Diagnose the visibility drop for project "$ARGUMENTS" (resolve via `get_all_projects` if needed). Compare the drop period against the period before it. Do not skip steps or diagnose from the aggregate alone.

1. `query_performance_report` dimensions=["topic", "date"] website_type="SELF" — which topics declined.
2. `query_performance_report` dimensions=["model", "date"] website_type="SELF" — which platforms declined.
3. `query_citation_report` dimensions=["date"] website_type="SELF" — were citations lost?
4. `query_sentiment_report` dimensions=["date"] website_type="SELF" — did sentiment turn?
5. `get_ai_answer_changes` — the exact prompts where answers changed; spot-check 2–3 with `get_prompt_answers`.
6. Check for configuration changes (prompts or competitors added/removed) — these shift denominators and can mimic a real drop.

Report: cause → evidence (numbers + period) → recommended fix, one block per cause. Declines first, no hyperbole.

View saved version →

---
name: weekly-geo-digest
description: Build the weekly GEO digest — visibility change vs last week, sentiment shift, biggest-moving prompts, top recommended actions. Use when the user asks for a weekly summary, digest, or "what moved this week".
---
Produce the weekly GEO digest for project "$ARGUMENTS" (call `get_all_projects` and ask if none is given). Use the last 7 days vs the 7 before.

1. `get_kpi_summary` with current_from/current_to and previous_from/previous_to — week-over-week visibility and mention deltas.
2. `query_sentiment_report` dimensions=["date"] website_type="SELF" — did sentiment shift?
3. `get_ai_answer_changes` — answers that appeared or disappeared this week (the biggest-moving prompts).
4. `get_actionables_dashboard` — recommended actions ranked by impact; pick the top three.

Format: four short sections in that order. Cite numbers with the period ("fell 3 pp from 41% to 38% WoW"), lead with declines, no hyperbole.