@wrenautomation/raven 0.2.1 → 0.3.0

package/README.md

@@ -10,9 +10,9 @@ Three calls, one record each:
 
 | Call | In | Out |
 |---|---|---|
-| `plan` | an ask (text, kind, what is known), proposals from people or other models, yield rows from earlier runs | a `Plan`: typed queries, each with `why`, `from` (ai or human) and `by` (model id, plan file or person), the sources it may use, the budget, and a snapshot of what raven was given |
-| `triage` | hits as a source shows them for free (title, channel, length, date, snippet, which sources returned it) | read or skip per hit, with why; the read budget caps it |
-| `review` | what a round came to (counts, found lines, notes) | the edge: `stop`, `again` (a new round from plan) or `back` (rerun one step), with why |
+| `plan` | an ask (text, kind, what is known, what the person says: `steer`, rendered first in every prompt), proposals from people or other models, yield rows from earlier runs | a `Plan`: typed queries, each with `why`, `from` (ai or human) and `by` (model id, plan file or person), the sources it may use, the budget, and a snapshot of what raven was given |
+| `triage` | hits as a source shows them for free (title, channel, length, date, snippet, which sources returned it) | read or skip per hit; the model writes a why on a read only (a skip's is fixed text); the read budget caps it |
+| `review` | what a round came to (counts, found lines, the extractor's verdict and gap per read, notes) | the edge: `stop`, `again` (a new round from plan) or `back` (rerun one step), with why |
 
 Ask kinds: `web`, `video`, `channel`, `person`, `site`. Each has a default plan file
 (prompts, sources, budget). Copy one, tune it, keep it beside your ledger: that is where a run
@@ -29,6 +29,7 @@ killed and the call fails with why.
 ```sh
 npm i @wrenautomation/raven
 npx raven plan "how do staffing firms win repeat clients" --known "referrals convert"
+npx raven plan "..." --steer "numbers only, skip vendor pages"   # what you say to the model, first in every prompt
 npx raven plan "..." --kind video --propose "cold email teardown" --by will      # yours merge with the model's
 npx raven triage --plan plan.json --hits hits.json
 npx raven review --round round.json

package/dist/app/cli.js

@@ -40,6 +40,7 @@ withModel(program
     .option("-k, --kind <kind>", `one of ${PLAN_KINDS.join(", ")}`, "web")
     .option("--site <host>", "keep every query on one host (kind site)")
     .option("--known <line>", "a fact already in hand; repeat", more, [])
+    .option("--steer <line>", "what you say to the model, rendered first; repeat", more, [])
     .option("--propose <terms>", "a query of your own; repeat", more, [])
     .option("--by <name>", "who proposed the --propose queries", who())
     .option("--plan-file <path>", "a tuned plan file (JSON) instead of the default")
@@ -47,7 +48,13 @@ withModel(program
     .option("--sources <names>", "narrow the plan file's sources, comma separated")
     .option("--round <n>", "round number", "1")).action(async (text, o) => {
     const kind = AskKind.parse(o.kind);
-    const ask = { text, kind, known: o.known, ...(o.site ? { site: o.site } : {}) };
+    const ask = {
+        text,
+        kind,
+        known: o.known,
+        steer: o.steer,
+        ...(o.site ? { site: o.site } : {}),
+    };
     const proposals = o.propose.length
         ? [
             {

package/dist/plan.js

@@ -5,7 +5,7 @@
  */
 import { fill } from "./model.js";
 import { defaultPlanFile } from "./plans/index.js";
-import { lines, renderYield } from "./render.js";
+import { lines, renderSteer, renderYield } from "./render.js";
 import { z, } from "./types.js";
 const ProposalAnswer = z.object({
     queries: z.array(z.object({ terms: z.string(), why: z.string(), site: z.string().optional() })),
@@ -16,6 +16,7 @@ export function proposePrompt(ask, planFile, yieldRows, now) {
         year: String(now.getUTCFullYear()),
         ask: ask.text,
         site: ask.site ?? "",
+        steer: renderSteer(ask.steer),
         known: lines(ask.known),
         yield: renderYield(yieldRows),
     });

package/dist/plans/index.js

@@ -6,39 +6,42 @@ import { PlanFile as PlanFileShape } from "../types.js";
 const SHAPE_QUERIES = `Return ONLY a JSON object, no prose, exactly this shape:
 {"queries": [{"terms": "...", "why": "..."}]}`;
 const SHAPE_HITS = `Return ONLY a JSON object, no prose, exactly this shape:
-{"hits": [{"url": "...", "read": true, "why": "..."}]}`;
+{"hits": [{"url": "...", "read": true, "why": "..."}, {"url": "...", "read": false}]}`;
 const SHAPE_EDGE = `Return ONLY a JSON object, no prose, exactly this shape:
 {"next": "stop", "why": "..."}
 where "next" is "stop", "again" (a new round from plan) or "back" (rerun one step of this round, named in "to": "plan", "fetch", "triage", "read" or "extract").`;
 const CONTEXT = `Ask: {ask}
-Known already (look past these):
+{steer}Known already (look past these):
 {known}
 What yielded before (hits, reads, proposed, kept):
 {yield}`;
 const TRIAGE_RULES = `- One entry per hit, with its url as given. Mark "read": true for at most {n}.
 - Read a hit when what it shows for free (title, snippet, channel, length, date) says it holds what the ask needs and is not known already.
 - A hit returned by more than one source is a good sign, not a duplicate.
-- "why" is one short line a person reads to agree or not.`;
+- "why" on a read only: one short line a person reads to agree or not. A skip has no "why".`;
 const REVIEW = `You decide what happens after one round of research.
 
 ${SHAPE_EDGE}
 
 Rules:
 - "stop" when the found lines answer the ask, or when another round would likely add nothing.
-- "again" when the ask is still open and new queries could find more. Say what to look for next in "why".
+- "again" when the ask is still open and new queries could find more. The gaps say what is missing; say what to look for next in "why".
+- "again" too when the reads were mostly "thin" or "off": the queries found the wrong pages, so plan different ones.
 - "back" when one step of this round failed for a reason a rerun fixes (a source was down, reads timed out). Name the step in "to".
 - "why" is one short line a person reads to agree or not.
 
 Ask: {ask}
-Round {round} of at most {rounds}. Hits {hits}, reads {reads}, findings proposed {proposed}, kept {kept}.
+{steer}Round {round} of at most {rounds}. Hits {hits}, reads {reads}, findings proposed {proposed}, kept {kept}.
 Found:
 {found}
+Read (quality: rich, thin, off or empty; then the gap, what the ask still lacks):
+{verdicts}
 Notes:
 {notes}`;
 const web = {
     id: "raven/web",
     kind: "web",
-    version: 1,
+    version: 3,
     propose: `You plan web searches for one research ask.
 
 ${SHAPE_QUERIES}
@@ -69,7 +72,7 @@ Hits:
 const video = {
     id: "raven/video",
     kind: "video",
-    version: 1,
+    version: 3,
     propose: `You plan video searches for one research ask: talks, walkthroughs, interviews and reviews on YouTube and podcast feeds.
 
 ${SHAPE_QUERIES}
@@ -100,7 +103,7 @@ Hits:
 const channel = {
     id: "raven/channel",
     kind: "channel",
-    version: 1,
+    version: 3,
     propose: `You plan searches that find the people and channels worth following for one research ask: creators, newsletters, podcasts, forums.
 
 ${SHAPE_QUERIES}
@@ -129,7 +132,7 @@ Hits:
 const person = {
     id: "raven/person",
     kind: "person",
-    version: 1,
+    version: 3,
     propose: `You plan searches about one person or company for one research ask: what they do, what they said, what others say about them.
 
 ${SHAPE_QUERIES}
@@ -160,7 +163,7 @@ Hits:
 const site = {
     id: "raven/site",
     kind: "site",
-    version: 1,
+    version: 3,
     propose: `You plan searches inside one website for one research ask. Every query runs on {site} only.
 
 ${SHAPE_QUERIES}

package/dist/render.d.ts

@@ -1,6 +1,8 @@
 /** Text the prompts take: what is known, what yielded, what a hit shows for free. */
 import type { HitView, YieldRow } from "./types.js";
 export declare const lines: (xs: readonly string[], none?: string) => string;
+/** "The person says:" and their lines, or nothing: a prompt pays for the heading only when there is one. */
+export declare const renderSteer: (steer: readonly string[] | undefined) => string;
 /** One line per row, the keys it has first: `source=exa channel=... hits=12 reads=4 proposed=9 kept=3`. */
 export declare function renderYield(rows: readonly YieldRow[]): string;
 /** Numbered, one hit per line, only the fields it has. */

package/dist/render.js

@@ -1,4 +1,6 @@
 export const lines = (xs, none = "(nothing)") => xs.length ? xs.map((x) => `- ${x}`).join("\n") : none;
+/** "The person says:" and their lines, or nothing: a prompt pays for the heading only when there is one. */
+export const renderSteer = (steer) => steer?.length ? `The person says:\n${lines(steer)}\n` : "";
 /** One line per row, the keys it has first: `source=exa channel=... hits=12 reads=4 proposed=9 kept=3`. */
 export function renderYield(rows) {
     if (!rows.length)

package/dist/review.js

@@ -5,16 +5,18 @@
  */
 import { fill } from "./model.js";
 import { defaultPlanFile } from "./plans/index.js";
-import { lines } from "./render.js";
+import { lines, renderSteer } from "./render.js";
 import { Step, z } from "./types.js";
 const EdgeAnswer = z.object({
     next: z.enum(["stop", "again", "back"]),
     to: Step.optional(),
     why: z.string(),
 });
+const verdictLine = (v) => `${v.url}: ${v.quality}${v.gap ? `. ${v.gap}` : ""}`;
 export function reviewPrompt(round, planFile) {
     return fill(planFile.review, {
         ask: round.ask.text,
+        steer: renderSteer(round.ask.steer),
         round: String(round.round),
         rounds: String(round.budget.rounds),
         hits: String(round.hits),
@@ -23,6 +25,7 @@ export function reviewPrompt(round, planFile) {
         kept: String(round.kept),
         found: lines(round.found),
         notes: lines(round.notes, "(none)"),
+        verdicts: lines(round.verdicts.map(verdictLine), "(none)"),
     });
 }
 export async function review(round, model, options = {}) {

package/dist/triage.d.ts

@@ -1,10 +1,13 @@
 /**
  * The filter before the read: for each hit, read or skip, from what the hit shows for free.
  * Hits dedupe by url and keep every source that returned them. The model picks; the read budget
- * caps it; a hit the model did not name is a skip, written down as one.
+ * caps it; a hit the model did not name is a skip, written down as one. The model writes a why
+ * on a read only: a skip's why is fixed text, so the answer pays for the reads alone.
  */
 import { type Model } from "./model.js";
 import { HitView, type HitViewIn, type Plan, type PlanFile, type Triage, type YieldRow } from "./types.js";
+/** What a skip says: the model named it and passed. */
+export declare const SKIPPED = "skipped by the model";
 /** One hit per url, sources merged, order of first sight. */
 export declare function dedupeHits(hits: readonly HitViewIn[]): HitView[];
 export declare function triagePrompt(hits: readonly HitView[], plan: Plan, planFile: PlanFile, yieldRows: readonly YieldRow[]): string;

package/dist/triage.js

@@ -1,15 +1,18 @@
 /**
  * The filter before the read: for each hit, read or skip, from what the hit shows for free.
  * Hits dedupe by url and keep every source that returned them. The model picks; the read budget
- * caps it; a hit the model did not name is a skip, written down as one.
+ * caps it; a hit the model did not name is a skip, written down as one. The model writes a why
+ * on a read only: a skip's why is fixed text, so the answer pays for the reads alone.
  */
 import { fill } from "./model.js";
 import { defaultPlanFile } from "./plans/index.js";
-import { lines, renderHits, renderYield } from "./render.js";
+import { lines, renderHits, renderSteer, renderYield } from "./render.js";
 import { HitView, z, } from "./types.js";
 const TriageAnswer = z.object({
-    hits: z.array(z.object({ url: z.string(), read: z.boolean(), why: z.string() })),
+    hits: z.array(z.object({ url: z.string(), read: z.boolean(), why: z.string().nullable().optional() })),
 });
+/** What a skip says: the model named it and passed. */
+export const SKIPPED = "skipped by the model";
 /** One hit per url, sources merged, order of first sight. */
 export function dedupeHits(hits) {
     const byUrl = new Map();
@@ -35,6 +38,7 @@ export function triagePrompt(hits, plan, planFile, yieldRows) {
         n: String(plan.budget.reads),
         ask: plan.ask.text,
         site: plan.ask.site ?? "",
+        steer: renderSteer(plan.ask.steer),
         known: lines(plan.ask.known),
         yield: renderYield(yieldRows),
         hits: renderHits(hits),
@@ -55,11 +59,11 @@ export async function triage(hits, plan, model, options = {}) {
         if (!a)
             return { ...base, read: false, why: "not named by the model" };
         if (!a.read)
-            return { ...base, read: false, why: a.why };
+            return { ...base, read: false, why: a.why?.trim() || SKIPPED };
         if (reads >= plan.budget.reads)
             return { ...base, read: false, why: `over the read budget (${plan.budget.reads})` };
         reads++;
-        return { ...base, read: true, why: a.why };
+        return { ...base, read: true, why: a.why?.trim() || "read" };
     });
 }
 //# sourceMappingURL=triage.js.map

package/dist/types.d.ts

@@ -30,6 +30,7 @@ export declare const Ask: z.ZodObject<{
         site: "site";
     }>;
     known: z.ZodDefault<z.ZodArray<z.ZodString>>;
+    steer: z.ZodDefault<z.ZodArray<z.ZodString>>;
     site: z.ZodOptional<z.ZodString>;
 }, z.core.$strip>;
 export type Ask = z.infer<typeof Ask>;
@@ -146,6 +147,7 @@ export declare const Plan: z.ZodObject<{
             site: "site";
         }>;
         known: z.ZodDefault<z.ZodArray<z.ZodString>>;
+        steer: z.ZodDefault<z.ZodArray<z.ZodString>>;
         site: z.ZodOptional<z.ZodString>;
     }, z.core.$strip>;
     round: z.ZodNumber;
@@ -208,7 +210,19 @@ export declare const Triage: z.ZodObject<{
     by: z.ZodString;
 }, z.core.$strip>;
 export type Triage = z.infer<typeof Triage>;
-/** What one round came to, as counts and notes. The runner fills it; raven reads it. */
+/** The extractor's word on one read document: how much of the ask it served, what is still missing. */
+export declare const Verdict: z.ZodObject<{
+    url: z.ZodString;
+    quality: z.ZodEnum<{
+        rich: "rich";
+        thin: "thin";
+        off: "off";
+        empty: "empty";
+    }>;
+    gap: z.ZodOptional<z.ZodString>;
+}, z.core.$strip>;
+export type Verdict = z.infer<typeof Verdict>;
+/** What one round came to, as counts, verdicts and notes. The runner fills it; raven reads it. */
 export declare const Round: z.ZodObject<{
     ask: z.ZodObject<{
         text: z.ZodString;
@@ -220,6 +234,7 @@ export declare const Round: z.ZodObject<{
             site: "site";
         }>;
         known: z.ZodDefault<z.ZodArray<z.ZodString>>;
+        steer: z.ZodDefault<z.ZodArray<z.ZodString>>;
         site: z.ZodOptional<z.ZodString>;
     }, z.core.$strip>;
     round: z.ZodNumber;
@@ -234,6 +249,16 @@ export declare const Round: z.ZodObject<{
     kept: z.ZodNumber;
     found: z.ZodDefault<z.ZodArray<z.ZodString>>;
     notes: z.ZodDefault<z.ZodArray<z.ZodString>>;
+    verdicts: z.ZodDefault<z.ZodArray<z.ZodObject<{
+        url: z.ZodString;
+        quality: z.ZodEnum<{
+            rich: "rich";
+            thin: "thin";
+            off: "off";
+            empty: "empty";
+        }>;
+        gap: z.ZodOptional<z.ZodString>;
+    }, z.core.$strip>>>;
 }, z.core.$strip>;
 export type Round = z.infer<typeof Round>;
 export declare const Step: z.ZodEnum<{

package/dist/types.js

@@ -14,6 +14,8 @@ export const Ask = z.object({
     kind: AskKind,
     /** Facts already in hand, one line each: the planner looks past them. */
     known: z.array(z.string()).default([]),
+    /** What the person said along the way, one line each: every prompt renders it first. */
+    steer: z.array(z.string()).default([]),
     /** Only for kind `site`, or when a query must stay on one host. */
     site: z.string().optional(),
 });
@@ -103,11 +105,18 @@ export const HitView = z.object({
 export const Triage = z.object({
     url: z.string(),
     read: z.boolean(),
+    /** The model's line on a read; a skip's is fixed text, so the model writes none. */
     why: z.string(),
     from: From,
     by: z.string(),
 });
-/** What one round came to, as counts and notes. The runner fills it; raven reads it. */
+/** The extractor's word on one read document: how much of the ask it served, what is still missing. */
+export const Verdict = z.object({
+    url: z.string(),
+    quality: z.enum(["rich", "thin", "off", "empty"]),
+    gap: z.string().optional(),
+});
+/** What one round came to, as counts, verdicts and notes. The runner fills it; raven reads it. */
 export const Round = z.object({
     ask: Ask,
     round: z.number().int().positive(),
@@ -120,6 +129,8 @@ export const Round = z.object({
     found: z.array(z.string()).default([]),
     /** What went wrong or ran out, one line each. */
     notes: z.array(z.string()).default([]),
+    /** The extractor's verdict on each document read this round: the planner sees what the reads were worth. */
+    verdicts: z.array(Verdict).default([]),
 });
 export const Step = z.enum(["plan", "fetch", "triage", "read", "extract"]);
 /** The edge out of a round. `again` is a new round from `plan`; `back` reruns one step of this round. */

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@wrenautomation/raven",
-  "version": "0.2.1",
+  "version": "0.3.0",
   "description": "Ask in, plan out. The planner of Wren's research engine: typed queries with who proposed them and why, proposals merged into one plan under a plan file per ask kind, read-or-skip on hits, and the edge to take after a round. Pure functions and prompts; the model is injected.",
   "license": "PolyForm-Shield-1.0.0",
   "repository": {