--- name: "literary_agent_querying" description: "Querying literary agents for a novel: a story ledger that keeps the synopsis honest, a query letter built from the book itself, agent research and fit ranking from verified guidelines, and a submission tracker that stops double queries. Triggers on query letter, synopsis, agent list, comp titles, QueryTracker, submission guidelines, partial or full request, offer of representation, or querying a novel or memoir." metadata: { "includeInPrompt": true } --- # Literary Agent Querying ## Purpose Get a finished novel in front of the right agents, in the exact form each agent asked for, with no duplicate or clashing submissions. The work is four linked pieces: a story ledger that tells the truth about the book, a query letter and synopsis built from that ledger, a ranked agent list built from current guidelines, and a tracker that records every send. Skip one piece and the others drift. This skill covers fiction querying. Nonfiction is queried on a proposal plus sample chapters before the book is finished. Do not apply the fiction workflow to nonfiction without rebuilding the package around the proposal. ## Workflow 1. **Gate the manuscript.** Query fiction only when the manuscript is complete, revised, and polished. Confirm the word count against the range for its age category and genre in `references/fit-scoring-and-research.md`. A count far outside the range is a revision problem first and a querying problem second. Fix the book or be ready to explain the length in one plain sentence. 2. **Build the story ledger.** Fill in `references/query-and-synopsis-worksheets.md`, part 1, from the manuscript itself: protagonist, goal, obstacle, stakes, inciting incident, midpoint turn, climax choice, ending, and what the protagonist gets wrong at the start. Every claim in the ledger must point to a chapter. If a ledger line cannot be found in the book, the book or the line is wrong. The ledger is the single source the query and synopsis are drawn from, so the two documents cannot contradict each other or the novel. 3. **Draft the query letter from the ledger.** One page, roughly 250 to 350 words. Structure: personalization line tied to this agent, metadata line (title, age category, genre, word count, two comps), two or three short story paragraphs covering protagonist, goal, obstacle, and stakes, then a brief bio and a thank you. Rules that matter: - The story paragraphs stop before the ending. A query teases, it does not resolve. - Name the choice the protagonist faces and what it costs either way. Vague stakes read as no stakes. - Comps: two titles, same age category and genre, published within the last 3 to 5 years, successful but not blockbuster franchises. Name the specific shared trait for each comp, and include the year. - Bio is relevant credits, relevant life experience, or one plain sentence for a debut. No unrelated jobs, no apologies for being unpublished. 4. **Draft the synopsis from the ledger.** One to two pages, roughly 500 to 800 words, third person, present tense, whatever the novel itself uses. Cover the full arc in order: setup, inciting incident, rising complications, midpoint turn, climax, and the actual ending. Include major twists. A synopsis that hides the ending fails its job, which is to let the agent judge whether the structure holds. Keep subplots to one line each or cut them. Format and cut with the worksheet in `references/query-and-synopsis-worksheets.md`, part 3. 5. **Research agents and score fit before anyone is queried.** Use QueryTracker for open or closed status and guidelines, Manuscript Wish List and the agent's own posts for stated wants, Publishers Marketplace deal reports for what the agent has actually sold, and the acknowledgments of recent comparable books for who represented them. Score each candidate with the sheet in `references/fit-scoring-and-research.md`. Record for every agent: agency, open status and the date you checked it, exact materials requested, submission method, stated response window, and the source URL. Guidelines change and agents close. A fact older than the current batch gets rechecked on the agency site before sending. 6. **Check every candidate for fee red flags.** A legitimate agent earns commission when the author gets paid. Under section 8 of the Canon of Ethics of the Association of American Literary Agents, member agents may not charge reading fees for evaluating queries, proposals, partials, or fulls, and under section 9 they may not charge clients for editorial services to prepare a work for submission or condition representation on buying those services. Any upfront reading, evaluation, or required editing fee removes the agent from the list, whatever the pitch. Not every good agent is a member, so absence from the member list is not a flag. A fee is. 7. **Assemble a per agent package, never a generic one.** For each agent, list exactly what their guidelines ask for and send only that: query only, query plus 5 pages, query plus 10 pages, query plus first three chapters, query plus synopsis, or another stated combination. All of those variants are in live use across agencies. Sending a synopsis nobody asked for, or the full manuscript uninvited, marks the sender as someone who did not read the guidelines. 8. **Log before you send, then query in batches.** Enter every agent in the tracker first with `scripts/tracker.py`, then log each query. The tracker blocks a second open query to the same agent for the same book and blocks a query to a second agent at an agency that already holds an open submission from you. Query in batches of 5 to 10 well matched agents. If a batch brings no requests, revise the query, the opening pages, or the target list before the next batch spends the rest of the good names. Worked batches for two genres are in `references/worked-packages.md`. 9. **Handle responses by the book.** - Partial or full request: send exactly what was asked, in the format asked, promptly, with a short cover note and no new pitch. - Exclusive request on requested material: it is your call, but if you grant it, agree a time limit in writing, four weeks at the outside, and pause new queries on that book until it lapses. - Silence: treat the agent's stated policy as the rule. Where the policy says no response means no, close the entry after the window passes. Nudge only where guidelines allow it, only after the stated window, and mainly on requested material. The tracker `due` command lists what is past its window. - Offer of representation: do not accept on the call. Thank the agent, set a decision date 7 to 14 days out, have the full conversation about editorial vision, submission plan, communication, and the agency agreement, then notify every other agent holding your query, partial, or full and give them that same date. The tracker `offer` command builds that notify list. Never invent a competing offer, and never name the offering agent to the others. - After accepting: withdraw promptly and politely from every agent still holding material. ## Output Contract Deliver, in this order: - The story ledger, completed from the manuscript, with chapter pointers. - The query letter, under 350 words, plus a per agent personalization line for each target. - The synopsis, 500 to 800 words, ending included. - The ranked agent list: name, agency, fit score with reasons, open status with check date, exact materials requested, submission method, response window, source URL. - The tracker file with every planned send logged, and a batch plan naming batch 1 and what result would trigger a package revision before batch 2. - A flags section: any word count problem, weak comp, fee flag, closed agent, or guideline you could not verify, stated plainly instead of smoothed over. ## Operating Rules 1. Accuracy first. Guidelines, open status, response windows, and deal history are facts: verify each on the agency site, QueryTracker, Manuscript Wish List, or Publishers Marketplace at the time of the batch, and cite the source and check date in the list. Never invent an agent's wants, clients, or sales. 2. The query never reveals the ending. The synopsis always does. Mixing those up is the most common package error. 3. One open submission per agency at a time unless that agency's current written guidelines say otherwise. The tracker enforces this. Do not work around it. 4. Send only what was requested, in the format requested. No extra chapters, no attachments where the form wants pasted text, no full manuscript uninvited. 5. Comps prove a current market. Older than five years, a different age category, or a franchise level bestseller does not prove one, however much the writer loves it. 6. No agent who charges to read, evaluate, or prepare your work stays on the list. State the fee you found as the reason for removal. 7. Keep every send, response, and date in the tracker. Memory is not a tracking system, and an untracked query is how double queries happen. 8. Write the next book while querying. Say it in the plan, because a querying plan with no next project turns every rejection into a verdict on the writer instead of data about a package.