@agent-pattern-labs/leads-rig 0.1.3 → 0.1.4

package/.opencode/skills/lead-harness.md

@@ -41,10 +41,10 @@ Available commands:
 
 Local commands:
   npx public-leads crawl --input data/domains.tsv --out data/lead-results.json
-  npx public-leads pipeline --input data/domains.tsv --ingest --target-project /path/to/cold-agent-leads
+  npx public-leads pipeline --input data/domains.tsv --ingest
   npx public-leads validate --input data/lead-results.json
   npx public-leads manifest --input data/lead-results.json
-  npx public-leads ingest --input data/lead-results.json --target-project /path/to/cold-agent-leads
+  npx public-leads ingest --input data/lead-results.json
   npx public-leads verify
 ```
 

package/.opencode/skills/public-leads.md

@@ -41,10 +41,10 @@ Available commands:
 
 Local commands:
   npx public-leads crawl --input data/domains.tsv --out data/lead-results.json
-  npx public-leads pipeline --input data/domains.tsv --ingest --target-project /path/to/cold-agent-leads
+  npx public-leads pipeline --input data/domains.tsv --ingest
   npx public-leads validate --input data/lead-results.json
   npx public-leads manifest --input data/lead-results.json
-  npx public-leads ingest --input data/lead-results.json --target-project /path/to/cold-agent-leads
+  npx public-leads ingest --input data/lead-results.json
   npx public-leads verify
 ```
 

package/.pi/iso-route.md

@@ -0,0 +1,15 @@
+# iso-route — Pi notes
+
+Pi reads `.pi/settings.json` for the default provider/model and model cycling. It does not have native role-specific subagent binding, so role entries below are advisory for manual model switching or extension/package workflows.
+
+## Default
+
+- **anthropic / claude-sonnet-4-6**
+
+## Roles
+
+| Role | Provider | Model | Thinking | Fallback |
+| ---- | -------- | ----- | -------- | -------- |
+| `fast` | anthropic | `claude-haiku-4-5` | - | - |
+| `quality` | anthropic | `claude-opus-4-7` | high | - |
+| `minimal` | anthropic | `claude-haiku-4-5` | - | - |

package/.pi/prompts/lead-harness.md

@@ -39,10 +39,10 @@ Available commands:
 
 Local commands:
   npx public-leads crawl --input data/domains.tsv --out data/lead-results.json
-  npx public-leads pipeline --input data/domains.tsv --ingest --target-project /path/to/cold-agent-leads
+  npx public-leads pipeline --input data/domains.tsv --ingest
   npx public-leads validate --input data/lead-results.json
   npx public-leads manifest --input data/lead-results.json
-  npx public-leads ingest --input data/lead-results.json --target-project /path/to/cold-agent-leads
+  npx public-leads ingest --input data/lead-results.json
   npx public-leads verify
 ```
 

package/.pi/prompts/public-leads.md

@@ -39,10 +39,10 @@ Available commands:
 
 Local commands:
   npx public-leads crawl --input data/domains.tsv --out data/lead-results.json
-  npx public-leads pipeline --input data/domains.tsv --ingest --target-project /path/to/cold-agent-leads
+  npx public-leads pipeline --input data/domains.tsv --ingest
   npx public-leads validate --input data/lead-results.json
   npx public-leads manifest --input data/lead-results.json
-  npx public-leads ingest --input data/lead-results.json --target-project /path/to/cold-agent-leads
+  npx public-leads ingest --input data/lead-results.json
   npx public-leads verify
 ```
 

package/.pi/settings.json

@@ -0,0 +1,9 @@
+{
+  "defaultProvider": "anthropic",
+  "defaultModel": "claude-sonnet-4-6",
+  "enabledModels": [
+    "claude-sonnet-4-6",
+    "claude-haiku-4-5",
+    "claude-opus-4-7"
+  ]
+}

package/README.md

@@ -20,9 +20,9 @@ npm run smoke:iso
 
 npx -p @agent-pattern-labs/leads-rig create-public-leads-harness my-lead-project
 npx public-leads crawl --domains example.com --out data/lead-results.json
-npx public-leads pipeline --input data/domains.tsv --ingest --target-project /path/to/cold-agent-leads
+PUBLIC_LEADS_API=https://cold-agent-leads.example.com npx public-leads pipeline --input data/domains.tsv --ingest
 npx public-leads validate --input examples/sample-leads.json
-npx public-leads ingest --input examples/sample-leads.json --dry-run --out output/sample-ingest.json
+PUBLIC_LEADS_API=https://cold-agent-leads.example.com npx public-leads ingest --input examples/sample-leads.json --dry-run --out output/sample-ingest.json
 batch/batch-runner.sh --help
 ```
 
@@ -32,12 +32,27 @@ The default reference client posts to the Cold Agent Leads ingest API:
 
 ```http
 POST /api/lead-ingests
-Authorization: Bearer $ADMIN_API_TOKEN
+Authorization: Bearer $PUBLIC_LEADS_API_TOKEN
 X-Admin-Email: admin@example.com
 Content-Type: application/json
 ```
 
-All endpoint details are configurable through `config/profile.yml`, CLI flags, environment variables, or `--target-project /path/to/cold-agent-leads` for reading that app's `.env` at runtime. Payloads contain `jobId`, `domains`, `leads`, `results`, and `errors`, with the schema defined in `templates/lead-schema.json`.
+For upstream ingest, set `PUBLIC_LEADS_API`, `PUBLIC_LEADS_API_TOKEN`, and `PUBLIC_LEADS_OPERATOR_EMAIL`; no local Postgres or local Cold Agent Leads server is required. Endpoint details are configurable through environment variables, `config/profile.yml`, or CLI flags. `--target-project /path/to/cold-agent-leads` remains available only as compatibility for reading a local app `.env`. Payloads contain `jobId`, `domains`, `leads`, `results`, and `errors`, with the schema defined in `templates/lead-schema.json`.
+
+Run from a Cold Agent Leads checkout without installing local package dependencies:
+
+```bash
+cd /Users/charlie/Razroo/cold-agent-leads
+export PUBLIC_LEADS_API=https://cold-agent-leads.example.com
+export PUBLIC_LEADS_API_TOKEN=...
+export PUBLIC_LEADS_OPERATOR_EMAIL=admin@example.com
+npx -p @agent-pattern-labs/leads-rig public-leads pipeline \
+  --domains example.com \
+  --ingest \
+  --out data/lead-results.json \
+  --manifest data/lead-manifest.json \
+  --ingest-out data/ingest-response.json
+```
 
 Legacy aliases `lead-harness` and `create-leads-harness` remain available for compatibility.
 
@@ -53,7 +68,7 @@ gh run watch
 You can also publish by creating a GitHub release for the package version:
 
 ```bash
-gh release create v0.1.3 --title v0.1.3 --generate-notes
+gh release create v0.1.4 --title v0.1.4 --generate-notes
 ```
 
 The package version must not already exist on npm.

package/bin/create-leads-harness.mjs

@@ -51,7 +51,7 @@ write('package.json', JSON.stringify({
     'update-harness': 'npm install @agent-pattern-labs/leads-rig@latest && public-leads sync',
   },
   dependencies: {
-    '@agent-pattern-labs/leads-rig': '^0.1.3',
+    '@agent-pattern-labs/leads-rig': '^0.1.4',
   },
   engines: {
     node: '>=20.6.0',
@@ -143,7 +143,7 @@ npm run verify
 opencode
 \`\`\`
 
-Update \`config/profile.yml\` with your ingest API settings and add domains to \`data/domains.tsv\` or \`data/pipeline.md\`.
+Set \`PUBLIC_LEADS_API\`, \`PUBLIC_LEADS_API_TOKEN\`, and \`PUBLIC_LEADS_OPERATOR_EMAIL\` for upstream ingest. Add domains to \`data/domains.tsv\` or \`data/pipeline.md\`.
 `);
 
 console.log('\nNext: npm install, then edit config/profile.yml and data/domains.tsv.\n');

package/config/profile.example.yml

@@ -2,16 +2,19 @@
 # Prefer environment variables for secrets in shared terminals.
 
 api:
-  # Cold Agent Leads local API default. Set target_project to a local
-  # cold-agent-leads checkout to read ADMIN_API_TOKEN/ADMIN_EMAILS from .env.
-  base_url: "http://localhost:8080"
+  # Set the deployed Cold Agent Leads API base URL through PUBLIC_LEADS_API,
+  # or put it here for a non-shared local profile. Example:
+  # base_url: "https://cold-agent-leads.example.com"
+  base_url: ""
   ingest_path: "/api/lead-ingests"
+  # Optional compatibility only: set when reading ADMIN_API_TOKEN/ADMIN_EMAILS
+  # from a local cold-agent-leads checkout. Not needed for upstream ingest.
   target_project: ""
-  operator_email: "admin@example.com"
+  operator_email: ""
   operator_email_header: "X-Admin-Email"
   auth_header: "Authorization"
   auth_scheme: "Bearer"
-  auth_token_env: "ADMIN_API_TOKEN"
+  auth_token_env: "PUBLIC_LEADS_API_TOKEN"
 
 crawl:
   max_pages: 10

package/docs/CONSTRUCTION.md

@@ -37,4 +37,4 @@
 6. Run `npm audit`.
 7. Run `npm pack` and confirm generated runtime surfaces are included.
 8. Scaffold a clean consumer project and test install-time sync from the packed package or source tree.
-9. Publish through GitHub Actions with `gh workflow run publish.yml --ref main`, or create a published release with `gh release create v0.1.3 --title v0.1.3 --generate-notes`.
+9. Publish through GitHub Actions with `gh workflow run publish.yml --ref main`, or create a published release with `gh release create v0.1.4 --title v0.1.4 --generate-notes`.

package/docs/SETUP.md

@@ -32,7 +32,9 @@ Then edit:
 Use environment variables for secrets:
 
 ```bash
-export ADMIN_API_TOKEN=...
+export PUBLIC_LEADS_API=https://cold-agent-leads.example.com
+export PUBLIC_LEADS_API_TOKEN=...
+export PUBLIC_LEADS_OPERATOR_EMAIL=admin@example.com
 ```
 
 ## Validate A Payload
@@ -51,27 +53,43 @@ npx public-leads pipeline --input data/domains.tsv --out data/lead-results.json
 ## Dry-Run Ingest
 
 ```bash
-npx public-leads ingest --input examples/sample-leads.json --dry-run --out output/sample-ingest.json
+PUBLIC_LEADS_API=https://cold-agent-leads.example.com npx public-leads ingest --input examples/sample-leads.json --dry-run --out output/sample-ingest.json
 ```
 
 ## Live Ingest
 
 ```bash
-npx public-leads pipeline --input data/domains.tsv --ingest --target-project /path/to/cold-agent-leads
-npx public-leads ingest --input data/lead-results.json --target-project /path/to/cold-agent-leads
+npx public-leads pipeline --input data/domains.tsv --ingest
+npx public-leads ingest --input data/lead-results.json
+```
+
+Run from `/Users/charlie/Razroo/cold-agent-leads` without local Postgres or a local API:
+
+```bash
+cd /Users/charlie/Razroo/cold-agent-leads
+export PUBLIC_LEADS_API=https://cold-agent-leads.example.com
+export PUBLIC_LEADS_API_TOKEN=...
+export PUBLIC_LEADS_OPERATOR_EMAIL=admin@example.com
+npx -p @agent-pattern-labs/leads-rig public-leads pipeline \
+  --domains example.com \
+  --ingest \
+  --out data/lead-results.json \
+  --manifest data/lead-manifest.json \
+  --ingest-out data/ingest-response.json
 ```
 
 Required request settings come from flags, environment, or `config/profile.yml`:
 
 - `api.base_url`
 - `api.ingest_path`
-- `api.target_project`
 - `api.operator_email`
 - `api.operator_email_header`
 - `api.auth_header`
 - `api.auth_scheme`
 - `api.auth_token_env`
 
+Use `api.target_project` only when intentionally reading credentials from a local Cold Agent Leads checkout. Upstream ingest should use `PUBLIC_LEADS_API`, `PUBLIC_LEADS_API_TOKEN`, and `PUBLIC_LEADS_OPERATOR_EMAIL`.
+
 ## Batch
 
 Create `batch/batch-input.tsv`:

package/iso/commands/lead-harness.md

@@ -44,10 +44,10 @@ Available commands:
 
 Local commands:
   npx public-leads crawl --input data/domains.tsv --out data/lead-results.json
-  npx public-leads pipeline --input data/domains.tsv --ingest --target-project /path/to/cold-agent-leads
+  npx public-leads pipeline --input data/domains.tsv --ingest
   npx public-leads validate --input data/lead-results.json
   npx public-leads manifest --input data/lead-results.json
-  npx public-leads ingest --input data/lead-results.json --target-project /path/to/cold-agent-leads
+  npx public-leads ingest --input data/lead-results.json
   npx public-leads verify
 ```
 

package/iso/commands/public-leads.md

@@ -44,10 +44,10 @@ Available commands:
 
 Local commands:
   npx public-leads crawl --input data/domains.tsv --out data/lead-results.json
-  npx public-leads pipeline --input data/domains.tsv --ingest --target-project /path/to/cold-agent-leads
+  npx public-leads pipeline --input data/domains.tsv --ingest
   npx public-leads validate --input data/lead-results.json
   npx public-leads manifest --input data/lead-results.json
-  npx public-leads ingest --input data/lead-results.json --target-project /path/to/cold-agent-leads
+  npx public-leads ingest --input data/lead-results.json
   npx public-leads verify
 ```
 

package/lib/leadharness-crawler.mjs

@@ -1,7 +1,7 @@
 import { createHash } from 'crypto';
 import { lookup, resolveMx } from 'dns/promises';
 import { setTimeout as sleep } from 'timers/promises';
-import { cleanDomains, normalizeDomain } from './leadharness-leads.mjs';
+import { cleanDomains, dedupeLeadRecords, mergeLeadRecords, normalizeDomain } from './leadharness-leads.mjs';
 
 const DEFAULT_MAX_PAGES = 10;
 const HARD_MAX_PAGES = 25;
@@ -51,16 +51,14 @@ export async function crawlDomains(inputs, options = {}) {
       for (const lead of result.leads) {
         const key = leadKey(lead);
         const current = leadsByKey.get(key);
-        if (!current || isBetterLead(lead, current)) {
-          leadsByKey.set(key, lead);
-        }
+        leadsByKey.set(key, current ? mergeLeadRecords(current, lead) : lead);
       }
     } catch (error) {
       errors.push(`${target.domain}: ${error instanceof Error ? error.message : String(error)}`);
     }
   }
 
-  const leads = [...leadsByKey.values()].sort(sortLead);
+  const leads = dedupeLeadRecords([...leadsByKey.values()]).sort(sortLead);
   const domains = cleanDomains([
     ...targets.map((target) => target.domain),
     ...results.map((result) => result.domain),
@@ -173,9 +171,7 @@ async function crawlDomainTarget(target, options) {
       if (!isRelevantLead(lead, { minConfidence, includeBlocked: options.includeBlocked })) continue;
       const key = leadKey(lead);
       const current = leadsByKey.get(key);
-      if (!current || isBetterLead(lead, current)) {
-        leadsByKey.set(key, lead);
-      }
+      leadsByKey.set(key, current ? mergeLeadRecords(current, lead) : lead);
     }
 
     if (page.hasForm && isContactLikeURL(page.url)) {
@@ -183,9 +179,7 @@ async function crawlDomainTarget(target, options) {
       if (isRelevantLead(lead, { minConfidence, includeBlocked: options.includeBlocked })) {
         const key = leadKey(lead);
         const current = leadsByKey.get(key);
-        if (!current || isBetterLead(lead, current)) {
-          leadsByKey.set(key, lead);
-        }
+        leadsByKey.set(key, current ? mergeLeadRecords(current, lead) : lead);
       }
     }
 
@@ -199,7 +193,7 @@ async function crawlDomainTarget(target, options) {
     await sleep(boundedInteger(options.delayMs, DEFAULT_DELAY_MS, 0, 10_000));
   }
 
-  const leads = [...leadsByKey.values()].sort(sortLead);
+  const leads = dedupeLeadRecords([...leadsByKey.values()]).sort(sortLead);
   const completedAt = new Date().toISOString();
   if (pages.length === 0) {
     warnings.push(`no pages could be crawled for ${target.domain}`);
@@ -502,6 +496,11 @@ async function buildEmailLead(email, companyDomain, websiteUrl, page) {
     emailType,
     sourceUrl,
     sourceLabel: sourceLabel(sourceUrl),
+    sources: [{
+      url: sourceUrl,
+      label: sourceLabel(sourceUrl),
+      evidence: snippet,
+    }],
     evidence: snippet,
     extractionMethod: 'agentic_harness_public_page',
     verificationStatus: emailType === 'blocked' ? 'blocked' : verificationStatus,
@@ -533,6 +532,11 @@ function buildContactPath(domain, websiteUrl, page) {
     emailType: 'contact_path',
     sourceUrl,
     sourceLabel: sourceLabel(sourceUrl),
+    sources: [{
+      url: sourceUrl,
+      label: sourceLabel(sourceUrl),
+      evidence: contactPathEvidence(page),
+    }],
     evidence: contactPathEvidence(page),
     extractionMethod: 'agentic_harness_contact_form',
     verificationStatus: 'not_applicable',

package/lib/leadharness-ingest.mjs

@@ -14,10 +14,18 @@ export function resolveIngestConfig(opts = {}, profile = loadProfileConfig()) {
   const configuredOperatorEmail = opts.operatorEmail || env.PUBLIC_LEADS_OPERATOR_EMAIL || env.LEAD_HARNESS_OPERATOR_EMAIL || profile.operatorEmail;
   const adminEmails = splitList(env.ADMIN_EMAILS);
   const operatorEmail = configuredOperatorEmail || legacyOperatorEmail || adminEmails[0] || '';
-  const tokenEnv = opts.tokenEnv || profile.authTokenEnv || profile.adminTokenEnv || 'ADMIN_API_TOKEN';
+  const tokenEnv = opts.tokenEnv
+    || profile.authTokenEnv
+    || profile.adminTokenEnv
+    || (env.PUBLIC_LEADS_API_TOKEN ? 'PUBLIC_LEADS_API_TOKEN' : '')
+    || (env.LEAD_HARNESS_API_TOKEN ? 'LEAD_HARNESS_API_TOKEN' : '')
+    || 'PUBLIC_LEADS_API_TOKEN';
+  const baseUrl = opts.api || env.PUBLIC_LEADS_API || env.LEAD_HARNESS_API || profile.baseUrl || '';
 
   return {
-    endpoint: `${stripTrailingSlash(opts.api || env.PUBLIC_LEADS_API || env.LEAD_HARNESS_API || profile.baseUrl || 'http://localhost:8080')}${normalizeEndpointPath(opts.ingestPath || env.PUBLIC_LEADS_INGEST_PATH || env.LEAD_HARNESS_INGEST_PATH || profile.ingestPath || '/api/lead-ingests')}`,
+    endpoint: baseUrl
+      ? `${stripTrailingSlash(baseUrl)}${normalizeEndpointPath(opts.ingestPath || env.PUBLIC_LEADS_INGEST_PATH || env.LEAD_HARNESS_INGEST_PATH || profile.ingestPath || '/api/lead-ingests')}`
+      : '',
     operatorEmail,
     operatorEmailHeader: opts.operatorEmailHeader
       || opts.adminEmailHeader
@@ -25,8 +33,7 @@ export function resolveIngestConfig(opts = {}, profile = loadProfileConfig()) {
       || env.LEAD_HARNESS_OPERATOR_EMAIL_HEADER
       || profile.operatorEmailHeader
       || profile.adminEmailHeader
-      || (legacyOperatorEmail || adminEmails.length > 0 ? 'X-Admin-Email' : '')
-      || (operatorEmail ? 'X-Operator-Email' : ''),
+      || 'X-Admin-Email',
     authHeader: opts.authHeader || env.PUBLIC_LEADS_AUTH_HEADER || env.LEAD_HARNESS_AUTH_HEADER || profile.authHeader || 'Authorization',
     authScheme: opts.authScheme || env.PUBLIC_LEADS_AUTH_SCHEME || env.LEAD_HARNESS_AUTH_SCHEME || profile.authScheme || 'Bearer',
     tokenEnv,
@@ -56,6 +63,10 @@ export async function ingestPayload(payload, opts = {}) {
     errors: payload.errors,
   };
 
+  if (!config.endpoint) {
+    throw new Error('ingest API base URL is required (--api, $PUBLIC_LEADS_API, $LEAD_HARNESS_API, or api.base_url)');
+  }
+
   if (opts.dryRun) {
     const output = {
       status: 'DRY RUN',
@@ -69,10 +80,16 @@ export async function ingestPayload(payload, opts = {}) {
   }
 
   if (config.operatorEmailHeader && !config.operatorEmail) {
-    throw new Error('operator email is required when an operator email header is configured');
+    throw new Error('operator email is required when an operator email header is configured (--operator-email, $PUBLIC_LEADS_OPERATOR_EMAIL, or api.operator_email)');
   }
   if (!config.token) {
-    throw new Error(`API token is required (--token or $${config.tokenEnv})`);
+    const tokenHints = [...new Set([
+      `$${config.tokenEnv}`,
+      '$PUBLIC_LEADS_API_TOKEN',
+      '$LEAD_HARNESS_API_TOKEN',
+      '$ADMIN_API_TOKEN',
+    ])];
+    throw new Error(`API token is required (--token, ${tokenHints.slice(0, -1).join(', ')}, or ${tokenHints.at(-1)})`);
   }
 
   const headers = {

package/lib/leadharness-leads.mjs

@@ -94,7 +94,7 @@ export function normalizePayload(value, { now = new Date().toISOString(), source
     throw new Error('lead artifact must be a JSON array, JSONL records, or an object with a leads array');
   }
 
-  const leads = payload.leads.map((lead) => normalizeLead(lead, { now }));
+  const leads = dedupeLeadRecords(payload.leads.map((lead) => normalizeLead(lead, { now })));
   const results = Array.isArray(payload.results)
     ? payload.results.map((result) => normalizeResult(result, { now }))
     : [];
@@ -146,6 +146,11 @@ export function normalizeLead(input, { now = new Date().toISOString() } = {}) {
     emailType,
     sourceUrl,
     sourceLabel: stringValue(input.sourceLabel) || sourceUrl,
+    sources: normalizeLeadSources(input.sources, {
+      sourceUrl,
+      sourceLabel: stringValue(input.sourceLabel) || sourceUrl,
+      evidence: compactWhitespace(stringValue(input.evidence)),
+    }),
     evidence: compactWhitespace(stringValue(input.evidence)),
     extractionMethod: stringValue(input.extractionMethod) || 'agentic_harness',
     verificationStatus,
@@ -158,7 +163,7 @@ export function normalizeLead(input, { now = new Date().toISOString() } = {}) {
   };
 
   if (!lead.id) lead.id = stableLeadID(lead);
-  return lead;
+  return normalizeLeadRecord(lead);
 }
 
 export function normalizeResult(input, { now = new Date().toISOString() } = {}) {
@@ -169,7 +174,9 @@ export function normalizeResult(input, { now = new Date().toISOString() } = {})
   return {
     domain,
     websiteUrl: stringValue(input.websiteUrl) || (domain ? `https://${domain}/` : ''),
-    leads: Array.isArray(input.leads) ? input.leads.map((lead) => normalizeLead(lead, { now })) : [],
+    leads: Array.isArray(input.leads)
+      ? dedupeLeadRecords(input.leads.map((lead) => normalizeLead(lead, { now })))
+      : [],
     pages: Array.isArray(input.pages) ? input.pages.map(normalizePageVisit) : [],
     warnings: normalizeWarnings(input.warnings),
     completedAt: normalizeDateTime(input.completedAt, now),
@@ -299,6 +306,19 @@ export function validateLead(lead, path, issues) {
   if (!Array.isArray(lead.warnings)) {
     issues.push(issue('error', `${path}.warnings`, 'invalid_warnings', 'warnings must be an array'));
   }
+  if (!Array.isArray(lead.sources)) {
+    issues.push(issue('error', `${path}.sources`, 'invalid_sources', 'sources must be an array'));
+  } else {
+    lead.sources.forEach((source, index) => {
+      if (!source || typeof source !== 'object' || Array.isArray(source)) {
+        issues.push(issue('error', `${path}.sources[${index}]`, 'invalid_source', 'sources entries must be objects'));
+        return;
+      }
+      if (!isHttpUrl(source.url)) {
+        issues.push(issue('error', `${path}.sources[${index}].url`, 'invalid_source_url', 'sources entries require an http(s) URL'));
+      }
+    });
+  }
   if (lead.foundAt && Number.isNaN(Date.parse(lead.foundAt))) {
     issues.push(issue('error', `${path}.foundAt`, 'invalid_found_at', 'foundAt must be ISO datetime'));
   }
@@ -447,6 +467,127 @@ export function domainFromUrl(value) {
   }
 }
 
+export function dedupeLeadRecords(leads) {
+  const byKey = new Map();
+  for (const rawLead of leads || []) {
+    const lead = normalizeLeadRecord(rawLead);
+    const key = stableLeadKey(lead);
+    const current = byKey.get(key);
+    byKey.set(key, current ? mergeLeadRecords(current, lead) : lead);
+  }
+  return [...byKey.values()];
+}
+
+function normalizeLeadRecord(lead) {
+  const sources = normalizeLeadSources(lead?.sources, {
+    sourceUrl: lead?.sourceUrl,
+    sourceLabel: lead?.sourceLabel,
+    evidence: lead?.evidence,
+  });
+  const normalized = {
+    ...lead,
+    sourceUrl: sources[0]?.url || stringValue(lead?.sourceUrl),
+    sourceLabel: sources[0]?.label || stringValue(lead?.sourceLabel),
+    sources,
+    evidence: sources[0]?.evidence || compactWhitespace(lead?.evidence),
+    warnings: normalizeWarnings(lead?.warnings),
+    socialUrls: normalizeUrlList(lead?.socialUrls),
+    contactUrls: normalizeUrlList(lead?.contactUrls),
+  };
+  normalized.id = stableLeadID(normalized);
+  return normalized;
+}
+
+export function mergeLeadRecords(current, candidate) {
+  const left = normalizeLeadRecord(current);
+  const right = normalizeLeadRecord(candidate);
+  const primary = leadQualityScore(right) > leadQualityScore(left) ? right : left;
+  const secondary = primary === right ? left : right;
+
+  const merged = {
+    ...primary,
+    company: primary.company || secondary.company,
+    domain: primary.domain || secondary.domain,
+    websiteUrl: primary.websiteUrl || secondary.websiteUrl,
+    contactName: primary.contactName || secondary.contactName,
+    title: primary.title || secondary.title,
+    email: primary.email || secondary.email,
+    emailType: primary.emailType || secondary.emailType,
+    extractionMethod: primary.extractionMethod || secondary.extractionMethod,
+    verificationStatus: verificationRank(secondary.verificationStatus) > verificationRank(primary.verificationStatus)
+      ? secondary.verificationStatus
+      : primary.verificationStatus,
+    confidence: Math.max(primary.confidence || 0, secondary.confidence || 0),
+    phone: primary.phone || secondary.phone,
+    socialUrls: normalizeUrlList([...(primary.socialUrls || []), ...(secondary.socialUrls || [])]),
+    contactUrls: normalizeUrlList([...(primary.contactUrls || []), ...(secondary.contactUrls || [])]),
+    warnings: normalizeWarnings([...(primary.warnings || []), ...(secondary.warnings || [])]),
+  };
+
+  merged.foundAt = earliestDateTime(primary.foundAt, secondary.foundAt);
+  if ((secondary.evidence || '').length > (merged.evidence || '').length) {
+    merged.evidence = secondary.evidence;
+  }
+
+  merged.sources = mergeLeadSources(merged.sourceUrl, left.sources, right.sources);
+  merged.sourceUrl = merged.sources[0]?.url || merged.sourceUrl;
+  merged.sourceLabel = merged.sources[0]?.label || merged.sourceLabel;
+  if (!merged.evidence) {
+    merged.evidence = merged.sources[0]?.evidence || '';
+  }
+  merged.id = stableLeadID(merged);
+  return normalizeLeadRecord(merged);
+}
+
+function normalizeLeadSources(value, fallback = {}) {
+  const byUrl = new Map();
+  const ordered = [];
+  const addSource = (source) => {
+    const url = stringValue(source?.url);
+    if (!url) return;
+
+    const entry = {
+      url,
+      label: stringValue(source?.label) || defaultSourceLabel(url),
+      evidence: compactWhitespace(source?.evidence),
+    };
+    const current = byUrl.get(url);
+    if (!current) {
+      byUrl.set(url, entry);
+      ordered.push(url);
+      return;
+    }
+    byUrl.set(url, {
+      url,
+      label: current.label || entry.label,
+      evidence: entry.evidence.length > current.evidence.length ? entry.evidence : current.evidence,
+    });
+  };
+
+  if (Array.isArray(value)) {
+    value.forEach(addSource);
+  }
+  addSource({
+    url: fallback.sourceUrl,
+    label: fallback.sourceLabel,
+    evidence: fallback.evidence,
+  });
+
+  const preferredUrl = stringValue(fallback.sourceUrl);
+  return ordered
+    .map((url) => byUrl.get(url))
+    .sort((left, right) => sourceSortValue(right, preferredUrl) - sourceSortValue(left, preferredUrl) || left.label.localeCompare(right.label) || left.url.localeCompare(right.url));
+}
+
+function mergeLeadSources(preferredUrl, ...lists) {
+  return normalizeLeadSources(lists.flat(), { sourceUrl: preferredUrl });
+}
+
+function sourceSortValue(source, preferredUrl) {
+  if (preferredUrl && source.url === preferredUrl) return 10_000;
+  return pagePriority(source.url || '') * 100 + Math.min((source.evidence || '').length, 99);
+}
+
 export function stableLeadID(lead) {
   const key = stableLeadKey(lead);
   if (!key.replace(/\|/g, '').trim()) return `agentic-${Date.now()}`;
@@ -469,6 +610,77 @@ function stableLeadKey(lead) {
   ].join('|').toLowerCase();
 }
 
+function leadQualityScore(lead) {
+  return (lead?.confidence || 0) * 1000
+    + verificationRank(lead?.verificationStatus) * 100
+    + pagePriority(lead?.sourceUrl || '') * 10
+    + (Array.isArray(lead?.sources) ? lead.sources.length : 0) * 5
+    + (lead?.phone ? 25 : 0)
+    + (Array.isArray(lead?.socialUrls) ? lead.socialUrls.length : 0) * 5
+    + (Array.isArray(lead?.contactUrls) ? lead.contactUrls.length : 0) * 5
+    + (lead?.contactName ? 20 : 0)
+    + (lead?.title ? 10 : 0);
+}
+
+function verificationRank(value) {
+  const status = stringValue(value).toLowerCase();
+  if (status === 'verified' || status === 'mx_verified') return 4;
+  if (status === 'unverified') return 3;
+  if (status === 'not_applicable') return 2;
+  if (status === 'unknown') return 1;
+  return 0;
+}
+
+function earliestDateTime(...values) {
+  const dates = values
+    .map((value) => stringValue(value))
+    .filter(Boolean)
+    .map((value) => new Date(value))
+    .filter((value) => !Number.isNaN(value.getTime()))
+    .sort((left, right) => left.getTime() - right.getTime());
+  return dates[0]?.toISOString() || new Date().toISOString();
+}
+
+function defaultSourceLabel(sourceUrl) {
+  const path = safePath(sourceUrl).replace(/^\/+|\/+$/g, '');
+  if (!path) return 'Home';
+  const last = path.split('/').filter(Boolean).pop() || path;
+  return last.replace(/[-_]+/g, ' ').replace(/\b\w/g, (char) => char.toUpperCase());
+}
+
+function safePath(value) {
+  try {
+    return new URL(value).pathname.toLowerCase();
+  } catch {
+    return '';
+  }
+}
+
+function pagePriority(link) {
+  const path = safePath(link);
+  const weighted = [
+    ['contact', 100],
+    ['about', 90],
+    ['team', 85],
+    ['people', 85],
+    ['leadership', 85],
+    ['staff', 80],
+    ['founder', 80],
+    ['press', 65],
+    ['media', 65],
+    ['blog', 55],
+    ['author', 55],
+    ['career', 45],
+    ['privacy', 25],
+    ['legal', 20],
+    ['impressum', 20],
+  ];
+  for (const [needle, score] of weighted) {
+    if (path.includes(needle)) return score;
+  }
+  return path === '/' ? 50 : 0;
+}
+
 export function fileHash(path) {
   return createHash('sha256').update(readFileSync(path)).digest('hex');
 }
@@ -507,7 +719,7 @@ function normalizeVerificationStatus(value, emailType) {
 
 function normalizeWarnings(value) {
   if (!Array.isArray(value)) return [];
-  return value.map((item) => stringValue(item)).filter(Boolean);
+  return [...new Set(value.map((item) => stringValue(item)).filter(Boolean))];
 }
 
 function normalizeUrlList(value) {

package/modes/_shared.md

@@ -45,6 +45,7 @@ Confidence should reflect source quality, person specificity, domain match, and
 ```bash
 npx public-leads validate --input data/lead-results.json
 npx public-leads manifest --input data/lead-results.json
-npx public-leads ingest --input data/lead-results.json --dry-run
+PUBLIC_LEADS_API=https://cold-agent-leads.example.com npx public-leads ingest --input data/lead-results.json --dry-run
+npx public-leads pipeline --input data/domains.tsv --ingest
 npx public-leads verify
 ```

package/modes/ingest.md

@@ -12,20 +12,23 @@ Use this when the user asks to store validated leads in a configured ingest API.
    ```bash
    npx public-leads manifest --input <artifact>
    ```
-3. Confirm ingest credentials and request settings are available through `config/profile.yml` and the configured environment variable. Do not print token values.
+3. Confirm upstream ingest settings are available through environment variables, flags, or `config/profile.yml`. Do not print token values.
+   - `PUBLIC_LEADS_API`
+   - `PUBLIC_LEADS_API_TOKEN`
+   - `PUBLIC_LEADS_OPERATOR_EMAIL`
 
 ## Dry Run
 
 Use dry-run before live ingest unless the user explicitly requested live submission:
 
 ```bash
-npx public-leads ingest --input <artifact> --dry-run --out data/ingest-response.json
+PUBLIC_LEADS_API=https://cold-agent-leads.example.com npx public-leads ingest --input <artifact> --dry-run --out data/ingest-response.json
 ```
 
 ## Live Ingest
 
 ```bash
-npx public-leads ingest --input <artifact> --target-project /path/to/cold-agent-leads --out data/ingest-response.json
+npx public-leads ingest --input <artifact> --out data/ingest-response.json
 ```
 
 ## Output

package/modes/pipeline.md

@@ -20,7 +20,7 @@ Read domains from the first available source:
 5. For browser-heavy batch work, use `batch/batch-runner.sh --parallel 2`.
 6. Validate every artifact with `npx public-leads validate`.
 7. Merge validated artifacts into a manifest with `npx public-leads manifest --input <artifact>`.
-8. Ingest with `npx public-leads pipeline --ingest --target-project /path/to/cold-agent-leads` only when upload is requested.
+8. Ingest with `npx public-leads pipeline --ingest` only when upload is requested and upstream API env vars are configured.
 
 ## Output
 

package/modes/setup.md

@@ -9,7 +9,7 @@ Use this when the user asks to configure or inspect the lead harness.
    - `data/domains.tsv`
    - `data/pipeline.md`
    - direct domains in the user request
-3. Confirm the ingest API settings:
+3. Confirm the upstream ingest API settings:
    - `api.base_url`
    - `api.ingest_path`
    - `api.operator_email`
@@ -17,7 +17,8 @@ Use this when the user asks to configure or inspect the lead harness.
    - `api.auth_header`
    - `api.auth_scheme`
    - `api.auth_token_env`
-   - `api.target_project` when using a local Cold Agent Leads checkout for `.env` credentials
+   - or `PUBLIC_LEADS_API`, `PUBLIC_LEADS_API_TOKEN`, and `PUBLIC_LEADS_OPERATOR_EMAIL`
+   - `api.target_project` only when intentionally reading credentials from a local Cold Agent Leads checkout
 4. Run `npx public-leads verify`.
 
 ## Output

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@agent-pattern-labs/leads-rig",
-  "version": "0.1.3",
+  "version": "0.1.4",
   "description": "Agentic public-web lead discovery harness with portable ingest artifacts",
   "type": "module",
   "author": "Razroo",
@@ -181,6 +181,6 @@
     "@razroo/iso": "^0.3.1",
     "@razroo/iso-eval": "^0.4.0",
     "@razroo/iso-harness": "^0.8.0",
-    "@razroo/iso-route": "^0.5.3"
+    "@razroo/iso-route": "^0.6.0"
   }
 }

package/scripts/ingest.mjs

@@ -9,7 +9,7 @@ Usage:
   public-leads ingest --input <file> [--api <base-url>] [--ingest-path </path>]
                       [--operator-email <email>] [--operator-email-header <header>]
                       [--auth-header <header>] [--auth-scheme <scheme>]
-                      [--token <token>] [--token-env ADMIN_API_TOKEN]
+                      [--token <token>] [--token-env PUBLIC_LEADS_API_TOKEN]
                       [--target-project /path/to/cold-agent-leads]
                       [--job-id <id>] [--out data/ingest-response.json] [--dry-run]
 
@@ -18,7 +18,7 @@ Defaults are read from config/profile.yml when present:
   api.auth_header, api.auth_scheme, api.auth_token_env, api.target_project
 
 Compatibility fallbacks are still accepted:
-  --admin-email, api.admin_email, api.admin_token_env, $PUBLIC_LEADS_API_TOKEN
+  --admin-email, api.admin_email, api.admin_token_env, $ADMIN_API_TOKEN, --target-project
 `;
 
 const opts = parseArgs(process.argv.slice(2));

package/scripts/pipeline.mjs

@@ -23,11 +23,11 @@ Usage:
                         [--manifest data/lead-manifest.json]
                         [--max-pages 10] [--min-confidence 30]
                         [--ingest | --upload] [--dry-run]
-                        [--target-project /path/to/cold-agent-leads]
+                        [--api https://cold-agent-leads.example.com]
 
 Examples:
   public-leads pipeline --domains example.com,example.org
-  public-leads pipeline --input data/domains.tsv --ingest --target-project /path/to/cold-agent-leads
+  PUBLIC_LEADS_API=https://cold-agent-leads.example.com public-leads pipeline --input data/domains.tsv --ingest
 `;
 
 const opts = parseArgs(process.argv.slice(2));