| Agent | Status | Last run | Host | Detail | |
|---|---|---|---|---|---|
| linkedin-icp-scraper | FAIL | 7h ago | LAPTOP-SA7P8OFV | reported failure — python exit 1 | |
| collision-watchdog | stale | 32d ago | LAPTOP-SA7P8OFV | stale, 32d since last run — 3 finding(s); worst=low: Browser agents scheduled close together | |
| agent-6-continuous-improvement | ok | 38h ago | LAPTOP-SA7P8OFV | scope=offer-only setup=skip offer=change copy=skip | |
| cold-email-prospector | ok | 16h ago | LAPTOP-SA7P8OFV | emitted 5 new leads, 29 prospects scanned | |
| linkedin-dm-stager | ok | 14h ago | LAPTOP-SA7P8OFV | reminder pushed; 8 lead(s) ready to stage | |
| linkedin-dm-watcher | ok | 2h ago | LAPTOP-SA7P8OFV | ||
| pain-signal-discovery | ok | 14h ago | LAPTOP-SA7P8OFV | ||
| salesnav-corporate-prospector | ok | 5h ago | LAPTOP-SA7P8OFV | 0 new leads, 0 mapped to pipeline; 899 companies tracked, 137 saturated |
Live from agent heartbeats (OneDrive-synced), pushed by push_agent_health.py. Status "FAIL" is self-reported by the agent; "stale" means no run in over 3 days. Click an agent to see its last 7 runs. The hardcoded automation board below is editorial and may lag this.
| When (CT) | Agent | What happened | |
|---|---|---|---|
| Aug 9, 17:22 | Salesnav Corporate Prospector | ran fine: 0 new leads, 0 mapped to pipeline; 899 companies tracked, 137 saturated | |
| Aug 9, 17:15 | Salesnav Extractor | unknown: 0 people -> C:\Users\info\OneDrive\Desktop\ClaudeProjects\CourseCreationConsulting\infra\agents\agent-15-salesnav-extractor\data\salesnav_extract_20260809_222223.csv | |
| Aug 9, 16:30 | LinkedIn DM Watcher | ran fine: new alerts this run: 0 (4m) | |
| Aug 9, 14:30 | LinkedIn ICP Scraper | FAILED: python exit 1 (39m) | |
| Aug 9, 13:01 | Salesnav Corporate Prospector | ran fine: 0 new leads, 0 mapped to pipeline; 899 companies tracked, 137 saturated | |
| Aug 9, 12:45 | Salesnav Extractor | unknown: 0 people -> C:\Users\info\OneDrive\Desktop\ClaudeProjects\CourseCreationConsulting\infra\agents\agent-15-salesnav-extractor\data\salesnav_extract_20260809_180107.csv | |
| Aug 9, 12:30 | LinkedIn DM Watcher | ran fine: new alerts this run: 0 (3m) | |
| Aug 9, 10:30 | LinkedIn ICP Scraper | FAILED: python exit 1 (30m) | |
| Aug 9, 09:00 | LinkedIn DM Stager | ran fine: reminder pushed; 8 lead(s) ready to stage | |
| Aug 9, 08:45 | Salesnav Extractor | unknown: 0 people -> C:\Users\info\OneDrive\Desktop\ClaudeProjects\CourseCreationConsulting\infra\agents\agent-15-salesnav-extractor\data\salesnav_extract_20260809_135440.csv | |
| Aug 9, 08:30 | LinkedIn DM Watcher | ran fine: new alerts this run: 0 (4m) | |
| Aug 9, 08:00 | Pain Signal Discovery | ran fine: 27m | |
| Aug 9, 06:45 | Cold Email Prospector | ran fine: emitted 5 leads to batch; 29 prospects to YT CSV (3m) | |
| Aug 9, 06:30 | LinkedIn ICP Scraper | ran fine: views_used=14/150 prospects=1 dm_queue=1 email_verified=0 (41m) | |
| Aug 8, 20:30 | LinkedIn DM Watcher | ran fine: new alerts this run: 0 (3m) | |
| Aug 8, 17:24 | Salesnav Corporate Prospector | ran fine: 0 new leads, 0 mapped to pipeline; 899 companies tracked, 137 saturated | |
| Aug 8, 17:15 | Salesnav Extractor | unknown: 0 people -> C:\Users\info\OneDrive\Desktop\ClaudeProjects\CourseCreationConsulting\infra\agents\agent-15-salesnav-extractor\data\salesnav_extract_20260808_222412.csv | |
| Aug 8, 16:30 | LinkedIn DM Watcher | ran fine: new alerts this run: 0 (4m) | |
| Aug 8, 14:30 | LinkedIn ICP Scraper | ran fine: views_used=48/150 prospects=0 dm_queue=0 email_verified=0 (40m) | |
| Aug 8, 12:49 | Salesnav Corporate Prospector | ran fine: 0 new leads, 0 mapped to pipeline; 899 companies tracked, 137 saturated | |
| Aug 8, 12:45 | Salesnav Extractor | unknown: 0 people -> C:\Users\info\OneDrive\Desktop\ClaudeProjects\CourseCreationConsulting\infra\agents\agent-15-salesnav-extractor\data\salesnav_extract_20260808_174933.csv | |
| Aug 8, 12:30 | LinkedIn DM Watcher | ran fine: new alerts this run: 0 (5m) | |
| Aug 8, 10:30 | LinkedIn ICP Scraper | ran fine: views_used=34/150 prospects=2 dm_queue=2 email_verified=0 (68m) | |
| Aug 8, 09:00 | LinkedIn DM Stager | ran fine: reminder pushed; 8 lead(s) ready to stage | |
| Aug 8, 08:45 | Salesnav Extractor | unknown: 0 people -> C:\Users\info\OneDrive\Desktop\ClaudeProjects\CourseCreationConsulting\infra\agents\agent-15-salesnav-extractor\data\salesnav_extract_20260808_135105.csv | |
| Aug 8, 08:30 | LinkedIn DM Watcher | ran fine: new alerts this run: 0 (4m) | |
| Aug 8, 08:10 | Agent 6 Continuous Improvement | ran fine: scope=offer-only setup=skip offer=change copy=skip | |
| Aug 8, 08:00 | Pain Signal Discovery | ran fine: 18m | |
| Aug 8, 06:45 | Cold Email Prospector | ran fine: emitted 10 leads to batch; 45 prospects to YT CSV (4m) | |
| Aug 8, 06:30 | LinkedIn ICP Scraper | ran fine: views_used=11/150 prospects=0 dm_queue=0 email_verified=0 (37m) | |
| Aug 7, 20:30 | LinkedIn DM Watcher | ran fine: new alerts this run: 0 (4m) | |
| Aug 7, 17:31 | Salesnav Corporate Prospector | ran fine: 0 new leads, 0 mapped to pipeline; 899 companies tracked, 137 saturated | |
| Aug 7, 17:15 | Salesnav Extractor | unknown: 0 people -> C:\Users\info\OneDrive\Desktop\ClaudeProjects\CourseCreationConsulting\infra\agents\agent-15-salesnav-extractor\data\salesnav_extract_20260807_223150.csv | |
| Aug 7, 16:30 | LinkedIn DM Watcher | ran fine: new alerts this run: 0 (3m) | |
| Aug 7, 14:30 | LinkedIn ICP Scraper | ran fine: views_used=46/150 prospects=7 dm_queue=7 email_verified=0 (61m) | |
| Aug 7, 13:00 | Salesnav Corporate Prospector | ran fine: 0 new leads, 0 mapped to pipeline; 899 companies tracked, 137 saturated | |
| Aug 7, 12:45 | Salesnav Extractor | unknown: 0 people -> C:\Users\info\OneDrive\Desktop\ClaudeProjects\CourseCreationConsulting\infra\agents\agent-15-salesnav-extractor\data\salesnav_extract_20260807_180007.csv | |
| Aug 7, 12:30 | LinkedIn DM Watcher | ran fine: new alerts this run: 0 (5m) | |
| Aug 7, 10:30 | LinkedIn ICP Scraper | ran fine: views_used=26/150 prospects=5 dm_queue=5 email_verified=0 (49m) | |
| Aug 7, 09:00 | LinkedIn DM Stager | ran fine: reminder pushed; 8 lead(s) ready to stage | |
| Aug 7, 08:45 | Salesnav Extractor | unknown: 0 people -> C:\Users\info\OneDrive\Desktop\ClaudeProjects\CourseCreationConsulting\infra\agents\agent-15-salesnav-extractor\data\salesnav_extract_20260807_135759.csv | |
| Aug 7, 08:30 | LinkedIn DM Watcher | ran fine: new alerts this run: 0 (3m) | |
| Aug 7, 08:00 | Pain Signal Discovery | ran fine: 23m | |
| Aug 7, 06:45 | Cold Email Prospector | ran fine: emitted 9 leads to batch; 39 prospects to YT CSV (3m) | |
| Aug 7, 06:30 | LinkedIn ICP Scraper | ran fine: views_used=8/150 prospects=1 dm_queue=1 email_verified=0 (30m) | |
| Aug 6, 20:30 | LinkedIn DM Watcher | ran fine: new alerts this run: 0 (6m) | |
| Aug 6, 17:15 | Salesnav Corporate Prospector | FAILED: scraper exit 99 | |
| Aug 6, 17:15 | Salesnav Extractor | unknown: started, no completion recorded (interrupted?) | |
| Aug 6, 16:30 | LinkedIn DM Watcher | FAILED: wrapper exception (2m) | |
| Aug 6, 14:30 | LinkedIn ICP Scraper | FAILED: wrapper exception (4m) | |
| Aug 6, 12:45 | Salesnav Corporate Prospector | FAILED: scraper exit 99 | |
| Aug 6, 12:45 | Salesnav Extractor | unknown: started, no completion recorded (interrupted?) | |
| Aug 6, 12:30 | LinkedIn DM Watcher | FAILED: wrapper exception (4m) | |
| Aug 6, 10:30 | LinkedIn ICP Scraper | FAILED: wrapper exception (4m) | |
| Aug 6, 09:00 | LinkedIn DM Stager | ran fine: reminder pushed; 8 lead(s) ready to stage | |
| Aug 6, 08:45 | Salesnav Extractor | unknown: started, no completion recorded (interrupted?) | |
| Aug 6, 08:30 | LinkedIn DM Watcher | FAILED: wrapper exception (4m) | |
| Aug 6, 08:00 | Pain Signal Discovery | FAILED: cdp unreachable on port 9223 (13m) | |
| Aug 6, 06:45 | Cold Email Prospector | ran fine: emitted 4 leads to batch; 31 prospects to YT CSV (5m) | |
| Aug 6, 06:30 | LinkedIn ICP Scraper | FAILED: wrapper exception (4m) | |
| Aug 5, 20:30 | LinkedIn DM Watcher | FAILED: wrapper exception (4m) | |
| Aug 5, 17:15 | Salesnav Corporate Prospector | FAILED: scraper exit 99 | |
| Aug 5, 17:15 | Salesnav Extractor | unknown: started, no completion recorded (interrupted?) | |
| Aug 5, 16:30 | LinkedIn DM Watcher | FAILED: wrapper exception (4m) | |
| Aug 5, 14:30 | LinkedIn ICP Scraper | FAILED: wrapper exception (4m) | |
| Aug 5, 12:45 | Salesnav Corporate Prospector | FAILED: scraper exit 99 | |
| Aug 5, 12:45 | Salesnav Extractor | unknown: started, no completion recorded (interrupted?) | |
| Aug 5, 12:30 | LinkedIn DM Watcher | FAILED: wrapper exception (5m) | |
| Aug 5, 10:30 | LinkedIn ICP Scraper | FAILED: wrapper exception (4m) | |
| Aug 5, 09:00 | LinkedIn DM Stager | ran fine: reminder pushed; 8 lead(s) ready to stage | |
| Aug 5, 08:45 | Salesnav Extractor | unknown: started, no completion recorded (interrupted?) | |
| Aug 5, 08:30 | LinkedIn DM Watcher | FAILED: wrapper exception (2m) | |
| Aug 5, 08:00 | Pain Signal Discovery | FAILED: cdp unreachable on port 9223 (14m) | |
| Aug 5, 06:45 | Cold Email Prospector | ran fine: emitted 7 leads to batch; 26 prospects to YT CSV (3m) | |
| Aug 5, 06:30 | LinkedIn ICP Scraper | FAILED: wrapper exception (4m) | |
| Aug 4, 20:30 | LinkedIn DM Watcher | FAILED: wrapper exception (4m) | |
| Aug 4, 17:15 | Salesnav Corporate Prospector | FAILED: scraper exit 99 | |
| Aug 4, 17:15 | Salesnav Extractor | unknown: started, no completion recorded (interrupted?) | |
| Aug 4, 16:30 | LinkedIn DM Watcher | FAILED: wrapper exception (4m) | |
| Aug 4, 14:30 | LinkedIn ICP Scraper | FAILED: wrapper exception (4m) | |
| Aug 4, 12:45 | Salesnav Corporate Prospector | FAILED: scraper exit 99 | |
| Aug 4, 12:45 | Salesnav Extractor | unknown: started, no completion recorded (interrupted?) | |
| Aug 4, 12:30 | LinkedIn DM Watcher | FAILED: wrapper exception (4m) | |
| Aug 4, 10:30 | LinkedIn ICP Scraper | FAILED: wrapper exception (3m) | |
| Aug 4, 09:00 | LinkedIn DM Stager | ran fine: reminder pushed; 8 lead(s) ready to stage | |
| Aug 4, 08:45 | Salesnav Extractor | unknown: started, no completion recorded (interrupted?) | |
| Aug 4, 08:30 | LinkedIn DM Watcher | FAILED: wrapper exception (2m) | |
| Aug 4, 08:00 | Pain Signal Discovery | FAILED: cdp unreachable on port 9223 (11m) | |
| Aug 4, 06:45 | Cold Email Prospector | ran fine: emitted 13 leads to batch; 58 prospects to YT CSV (6m) | |
| Aug 4, 06:30 | LinkedIn ICP Scraper | FAILED: wrapper exception (4m) | |
| Aug 3, 20:30 | LinkedIn DM Watcher | FAILED: wrapper exception (4m) | |
| Aug 3, 17:15 | Salesnav Corporate Prospector | FAILED: scraper exit 99 | |
| Aug 3, 17:15 | Salesnav Extractor | unknown: started, no completion recorded (interrupted?) | |
| Aug 3, 16:30 | LinkedIn DM Watcher | FAILED: wrapper exception (4m) | |
| Aug 3, 14:30 | LinkedIn ICP Scraper | ran fine: views_used=25/150 prospects=1 dm_queue=1 email_verified=0 (33m) | |
| Aug 3, 12:52 | Salesnav Corporate Prospector | ran fine: 0 new leads, 0 mapped to pipeline; 899 companies tracked, 137 saturated | |
| Aug 3, 12:45 | Salesnav Extractor | unknown: 0 people -> C:\Users\info\OneDrive\Desktop\ClaudeProjects\CourseCreationConsulting\infra\agents\agent-15-salesnav-extractor\data\salesnav_extract_20260803_175251.csv | |
| Aug 3, 12:30 | LinkedIn DM Watcher | ran fine: new alerts this run: 0 (5m) | |
| Aug 3, 10:30 | LinkedIn ICP Scraper | ran fine: views_used=19/150 prospects=1 dm_queue=1 email_verified=0 (49m) | |
| Aug 3, 09:00 | LinkedIn DM Stager | ran fine: reminder pushed; 8 lead(s) ready to stage | |
| Aug 3, 08:45 | Salesnav Extractor | unknown: 0 people -> C:\Users\info\OneDrive\Desktop\ClaudeProjects\CourseCreationConsulting\infra\agents\agent-15-salesnav-extractor\data\salesnav_extract_20260803_135116.csv | |
| Aug 3, 08:30 | LinkedIn DM Watcher | ran fine: new alerts this run: 0 (4m) | |
| Aug 3, 08:00 | Pain Signal Discovery | ran fine: 17m | |
| Aug 3, 06:45 | Cold Email Prospector | ran fine: emitted 11 leads to batch; 50 prospects to YT CSV (5m) | |
| Aug 3, 06:30 | LinkedIn ICP Scraper | ran fine: views_used=1/150 prospects=0 dm_queue=0 email_verified=0 (23m) |
Built from each agent's own run logs (OneDrive-synced), pushed by push_agent_runs.py, so it now captures every run — even agents that run several times a day. Click an agent to see just its last 7 runs. The Sales Nav corporate agent (Agent 15, salesnav-corporate-prospector) is semi-manual and has no run-log folder, so it shows its latest heartbeat only.
| Audience | Leads | Cold sent | Replies | Positive |
|---|---|---|---|---|
| Corporate | 876 | 1,724 | 0 | 0 |
| Creators | 290 | 509 | 7 | 2 |
Live snapshot pushed by push_cold_metrics.py. "Replies" counts genuine human replies (auto-acknowledgements excluded). Historical campaigns are tagged by audience; going forward, leads route to a standing corporate or creators campaign rather than a new version.
Sourced / verified / screened are rough editorial estimates (as of 6/29, across YouTube/GitHub creators + the Sales-Nav corporate pivot; agent 12 alone harvested ~276 corporate engineers, ~42 verified). Contacted / replied / positive / booked / closed are live (DB-fed). The top three stay a manual snapshot until the funnel is wired to the per-source digests.
Reference only (editorial, may lag). For live run status use the Agent health panel near the top of the page, which reads the heartbeats. This table describes what each agent does and where its output lands; the "last run" column is a hand-maintained snapshot (refreshed 2026-06-29) — trust the live health panel for actual run times.
| Agent | Host | Does | Cadence | Last run | Output → lane | |
|---|---|---|---|---|---|---|
| ColdEmail-RunAll pipeline | — | Inbox watcher (replies + phone push on positives) → sender (20/run, 30/box/day cap) → suppression → warmup → metrics push to this dashboard. The 03:00 run adds mailbox retention cleanup + DB snapshot. DISABLED / held until explicit cold-send go + ICANN verification. | 6x daily | held (disabled) | sends + replies → cold email lane | |
| CE-Warmup-Alienware | Alienware (LAPTOP-SA7P8OFV) | Warming only (ce.py warmup-once), no cold sender. Validated 2026-06-21 (sent 32, exit 0). | 6x daily | 2026-06-21, ok | warmup → cold email lane |
Migrated 2026-06-21: the host guard now pins sending to Alienware; Dell no-ops. The cold sender is currently held DISABLED on Alienware until the cold-send go + ICANN verification; warming continues via CE-Warmup-Alienware.
| Agent | Host | Does | Cadence | Last run | Output → lane | |
|---|---|---|---|---|---|---|
| Agent 16 · LinkedIn DM stager | Alienware (LAPTOP-SA7P8OFV) | LIVE 2026-06-29. Opens a tab per dm_ready lead with a question-first DM pre-typed (unsent); Maaz hits Send. Up to 8/day, US / high-income geo-gated, free-message only. 9AM ntfy reminder (cdb-maaz-leads). | daily 09:00 + on-demand | see health panel | staged DMs → LinkedIn DM lane | |
| Agent 5 · LinkedIn engineering-creator prospector | Alienware (LAPTOP-SA7P8OFV) | Finds engineering creators with 1000+ LinkedIn followers (US gate), read-only, 30-view/day cap. | daily 06:30 | see health panel | digest → LinkedIn DM lane | |
| Agent 15 · Sales Nav corporate sourcing (heartbeat: salesnav-corporate-prospector) | Alienware (LAPTOP-SA7P8OFV) | Pulls corporate engineers (10k+ headcount / US) straight from a Sales Navigator people search via a console collector (clean structured fields + company org-ids for auto-exclusion of saturated firms), then salesnav_aggregate.py maps domains -> verify (Reoon) -> backlog -> promote into engineers_v9 (held). Replaced the old video+OCR flow. There is no "Agent 12" folder — this is the corporate Sales Nav agent. | semi-manual: you run a Sales Nav session + aggregate; not an unattended daily chain | shows in the live panels only after the aggregate step runs (it writes the heartbeat) | verified leads → corporate cold lane | |
| Agents 10-11 · creator + podcast/conference sourcing | Alienware (LAPTOP-SA7P8OFV) | Agent 10 = GitHub engineering creators; Agent 11 = podcast/conference hosts. Built + live-validated 2026-06-23. Verify -> backlog -> promote. | ad-hoc / daily | see health panel | verified leads → cold + DM lanes | |
| Agent 7 · pain-signal discovery | Alienware (LAPTOP-SA7P8OFV) | Finds 10-20 named people publicly wanting to productize expertise (LinkedIn, Reddit, IndieHackers; phrase rotation) | daily 08:00 | see health panel | ranked digest → X intent lane + content ideas | |
| ColdEmail-DailyReport | Alienware (LAPTOP-SA7P8OFV) | Daily summary email of sends, replies, bounces | daily 08:30 | see health panel | email → inbox |
| Agent | Host | Does | Cadence | Last run | Output → lane | |
|---|---|---|---|---|---|---|
| Agent 6 · continuous improvement | Alienware (LAPTOP-SA7P8OFV) | Reads real outreach artifacts + the send database; outputs up to 3 literal copy/setup/offer diffs, never generic advice. Recreated on Alienware 2026-06-21; the prior Dell 'claude CLI exit 1' failure was environment-specific (the call works on Alienware). First Alienware run upcoming Saturday. | weekly Saturday | — (not yet run on Alienware) | diff memo → all lanes |
Agents 2, 3, 8 deleted (superseded by Agent 9); their Dell tasks unregistered 6/13.
| Agent | Host | Does | Cadence | Last run | Output → lane | |
|---|---|---|---|---|---|---|
| CLP-Day7-Batch | — | Day-7 batch task | disabled | — | per topology notes, intentional | |
| Agent 4 · LinkedIn DM watcher | Alienware (LAPTOP-SA7P8OFV) | LIVE as of 2026-06-22. Site-based: opens LinkedIn /messaging/ (Unread filter) + /notifications/, read-only, never sends. Pushes ntfy alerts (topic cdb-maaz-leads) for new DMs and accepted connection invites. | 4x daily (08:30 / 12:30 / 16:30 / 20:30) | 2026-06-22, ok | ntfy alert → LinkedIn DM lane |
| Agent | Host | Does | Cadence / state | |
|---|---|---|---|---|
| Agent 9 · cold-email prospector | Alienware (LAPTOP-SA7P8OFV) |
BUILT AND LIVE; scheduled daily 06:45 on Alienware (task "CLP Cold-Email Prospector"). Single mandate: ~100 engineering YouTube/GitHub creators per day. Lead finder — NEVER sends email. Sources: YouTube (PRIMARY, API key valid) + GitHub drawdown (secondary, finite).
Geo + recency fix (2026-06-29): the old US-only YouTube gate was starving the pool, so the gate is now a 44-country high-income allowlist plus recency_videos discovery (publishedAfter + English bias). Free YouTube Data API key ($0; 10,000 units/day, budget-capped at 3,000), search.list rotation + channels.list + website crawl + Reoon safe-only. YouTube yields few emails (links sit behind the About-page captcha), so it doubles as a DM-lane discovery machine banking "channel, no email" lists for the LinkedIn/manual DM lane. Location matcher uses word-boundary states/cities + the geo allowlist. Location-enrichment pass (GitHub profile re-fetch + personal-site crawl) recovers held verified-email leads. | SCHEDULED daily 06:45 (task "CLP Cold-Email Prospector", Alienware) |
| Agent | Host | Does | Starts when | |
|---|---|---|---|---|
| DeepSeek personalization drafting | — | Evidence-quoted one-liners per lead; Maaz reviews 100% of first 200 in batched sessions | with the next batch | |
| Listen Notes Pro podcast sourcing ($200/mo) | — | Education/business podcast hosts, listen-score filter, native email field (paid tier of agent 11's free sourcing) | gated: a 20-email manual test produces a booked call or paid Pre-Sell Sprint | |
| ColdSource-RunAll scheduler | — | Single daily sourcing pipeline unifying agents 9-12 (discover → filter → enrich → verify → dedupe) + morning digest | once the per-agent runs prove out |
Note: Agent 14 (Apollo corporate sourcing) was shelved — free API 403s and the web UI bot-blocks automation; Sales Navigator Core (via agent 12) is the corporate source instead.
Strategy. Two ICPs: YouTube creators (BUILD = no course yet; UPSELL = improve existing) and GitHub-sourced engineers with monetization signals. US-only for automated sends; UK/CA go to the manual lane. Volume ramps only on reply quality. Sourcing target once Agent 9 lands: 100 qualified/day capacity, throttle gated on proof.
| Automation | State | Schedule / next |
|---|---|---|
| Send + reply pipeline (watcher, sender, suppression, warmup, metrics push) | running | 6x daily: 03, 08, 11, 14, 17, 20 |
| Agent 9: GitHub + YouTube sourcing + monetization-signal layer | BUILT AND LIVE, scheduled daily 06:45 on Alienware. Geo + recency fixed 6/29: YouTube gate widened from US-only to a 44-country high-income allowlist + recency_videos discovery (was starving the pool). Yields DM-lane discovery (few emails behind the captcha) - every digest banks a "channel, no email" list for the LinkedIn/manual DM lane. | scheduled daily 06:45 (task "CLP Cold-Email Prospector"); also manual python agent9.py --target 100 |
| Personalization drafting (DeepSeek, evidence-quoted) + 100% founder review of first 200 | protocol agreed | starts with next batch |
| Email verification (Reoon, safe-only rule) | running | per batch |
Standing campaigns (no more versioning): corporate_engineers and creator_engineers. Current corporate copy is the v9 curriculum lead-magnet (148 drafts staged in ce.db, send held). Ground truth 2026-06-25: ~376 sent all-time, 4 real replies, 1 genuine positive (Tyler); deliverability verified good (mail-tester 10/10, SPF/DKIM/DMARC pass). Messaging: engineers sequence · youtube day-0
Strategy. The proof engine. Review a creator's FREE public asset as a working ID, send 3 specific strengths + 1 drop-off risk, offer the full audit on reply. Free-asset scope, 30-45 min cap each; the deep variant (buy course, rebuild module 1) reserved for top targets. Each teardown doubles as content and a case-study seed, and feeds the $1k Pre-Sell Sprint on resulting calls.
Drafts: message library. Lange + Reilly route via LinkedIn DM (Germany / no email). Watch: Adam Kae dropped for unverifiable data; never tear down on shaky evidence.
Strategy. Strong-fit leads the automated sender must not touch: UK (PECR) and Canada (CASL) compliance, plus personal catch-all domains Reoon cannot confirm. Hand-written one-offs from a real identity, date-logged.
UK: Liz Rice, Alex Ellis. CA: Andriy Burkov, Ali Hejazizo, Satoshi Tanda, Ajin Abraham. Catch-all: Kent C. Dodds, Steve Klabnik (or DM lane). Adapt the engineer day-0 bodies from the library.
Strategy (agreed in the Core Four plan). Follow-up lane, not a cold-volume engine. Q1 2026 was an enforcement wave (HeyReach banned in March; ~40% of cloud-tool accounts restricted), and this profile is also the warm-25 and referral channel, which makes it unreplaceable. So: zero automation tools on the profile, hand-written notes only, ~5/day ceiling, prioritize prospects who already replied or have no findable email.
| Agent (all read-only) | State | Schedule |
|---|---|---|
| Agent 3: LinkedIn lead finder (5 vetted/run, 30-view cap) | deleted 6/11 (old GRC ICP; superseded by Agent 9) | — |
| Agent 4: DM response watcher · LinkedIn DM watcher | LIVE as of 2026-06-22 on Alienware. Site-based: opens LinkedIn /messaging/ (Unread filter) + /notifications/, read-only, never sends. Pushes ntfy alerts (topic cdb-maaz-leads) for new DMs and accepted connection invites. | 4x daily (08:30 / 12:30 / 16:30 / 20:30) |
| Agent 5: LinkedIn engineering-creator prospector (1000+ followers, US gate) | active (ran 6/21 06:30 on Alienware) | daily 06:30 |
Suggested templates (draft, not yet through full curation): message library.
Strategy (from the four-model debate; DeepSeek's addition). Search X for people already saying it: "thinking about making a course", "my course flopped", "most courses are boring", "should I build a course". Hand-written replies/DMs only. Warmest possible cold channel because they raised the topic themselves. Also the routing for Dodds and Klabnik.
Suggested message shape: message library. Agent 7 (pain-signal discovery, daily 08:00, active) already surfaces some of these signals from LinkedIn/Reddit/IndieHackers; extend it to X queries as a later tweak.
Rule for the next 30 days: produce only what supports sales directly. The authority cluster waits for the first case study. Frame everything as revenue/retention/refunds; instructional design is the mechanism, not the headline.
| Piece | Type | When |
|---|---|---|
| How to pre-sell a course in 7 days (DM CTA) | carousel | this week |
| Teardown write-ups from outreach (anonymized or with permission) | post series | as teardowns send |
| Adult-learning principles series (spaced repetition, cognitive load, retrieval practice, worked examples) | carousel series | post-case-study |
| Most courses are boring because of one structural mistake | carousel | post-case-study |
| AI in instructional design: what it can and cannot replace | post | post-case-study |
| "I reviewed N creator courses as an ID" | post | only when teardown count makes N honest |
| Piece | Type | When |
|---|---|---|
| Pre-sell wedge landing page ("sell your course before you build it") + founder photo + TidyCal "Course Pre-Sell Call" scheduler | web page | DEPLOYED LIVE 6/29 (courselaunchpartner.com) |
| 7-step pre-sell checklist (download variant of the carousel) | 1-pager | next |
| Weekly beehiiv: one course-design teardown per issue | newsletter | gated: list exists |
| Piece | Type |
|---|---|
| Why your online course has a 10 percent completion rate (and the fix) | authority |
| How to create an online course: learning design structure, not platform setup | search (top query in niche) |
| Kajabi vs Teachable vs Skool through a learning-design lens | comparison (evergreen) |
| How course pre-selling works / why most courses fail (pre-sell fix) | explainer (existing board) |
| AI tools for course creation: what saves time vs needs rework | trust |
| Piece | Keyword | When |
|---|---|---|
| How much does it cost to hire someone to build your course | done for you course creation | next 30 days |
| Pre-sell an online course, step by step / launch without building | pre-sell online course | existing board, high |
| ID-lens platform comparison (twin of the video) | kajabi vs teachable vs skool | post-case-study |
| Pricing psychology tied to design scope | online course pricing | post-case-study |
Provenance shown per item. Nothing here sends itself; this is the single source of truth for what goes out.
Current standing copy is the v9 curriculum lead-magnet (overhauled 2026-06-27: free course-curriculum CTA, ~3rd-grade reading level, "second income" outcome, lead-course-designer / Fortune 500 authority). Its 148 per-lead drafts live in ce.db (send held); they are not pasted here to avoid drift from the database of record. The v6 sequence below is the prior creator-facing copy, kept for reference.
Day 0 · subject is per-person (sample: "a course for your pandas viewers?") · greeting/signature templated
Hi {first_name},
{one specific evidence-based observation, e.g.: Your Pandas tutorial repos have over a thousand stars each. That usually means the audience wants more than free videos.}
I help experts pre-sell a course before building it. Your audience pays up front. If not enough people buy, nothing gets built and you lose nothing. Do you have a course in mind that you are too busy to build?
Sarah
Course Launch Partner
Reply STOP to unsubscribe.
Day 3 · same thread, no subject
Hi {first_name},
Following up in case this slipped by. The idea still stands. I pre-sell a course to your audience before we build it, so you only make what people already paid for.
Worth a quick look? Even a no helps me plan.
Sarah
Reply STOP to unsubscribe.
Day 7 · same thread, referral close
Hi {first_name},
Last note from me on this. If the timing is off, no problem at all.
If you would rather point me to another creator, I pay 10% on any deal that closes from someone you send my way.
Sarah
Reply STOP to unsubscribe.
Decided 6/14: postal address in the footer is skipped for now per Maaz (US spam-law physical-address requirement flagged by both reviewers; accepted risk, revisit before scaling volume).
Day 0 shape (the one that got the Tyler reply)
Hi {first_name},
{personal congrats on a real event, e.g. the CBS Sunday Morning feature}. I help you pre-sell a course before we build it. You only make what people already paid for. Do you have one in mind that you're too busy to build? If not, I also pay 10% on any deal that closes from someone you send my way. Just say no if you aren't interested.
Best,
Sarah Smith
Tyler explainer (SENT 6/10, on hold since)
Hi Tyler, Happy to walk you through it. Here's how it works: 1. We pick the course. We look at your ideas together, plus what your viewers keep asking for. We pick the one people will pay for. 2. We sell it before we build it. A simple page goes out to your audience. People pay up front. No real orders, no course. 3. If enough people buy, we build it. Curriculum, production, platform, launch. You do interviews and review drafts. A few hours a month. If not enough people buy, they get refunds and you lose nothing. Format is up to you. Self-paced, live sessions, or both. Your part can be as small as you want. You can record everything yourself. Or just give us the content, and we can clone your voice and even create an AI version of you. You approve everything before it goes out. What courses do you have in mind? Easiest is a quick call. What day and time work best for you? Best, Sarah Smith
1 · Walter Haydock (StackAware) · info@stackaware.com or LinkedIn · send first
Subject: your 5-day ISO 42001 course Hi Walter, I took apart your free 5-day ISO 42001 email course as a working instructional designer. The day sequence is a real curriculum arc: why, comparison, scope, assess, controls. A lot of free email courses stop at framing; yours teaches. The Maven workshop structure is also strong, three sessions plus a project plus a 1:1. One gap: every day teaches about ISO 42001, but no day has the reader produce anything. One small artifact per day (day 3: a one-line AIMS scope statement) turns readers into people who have already started implementing, and that makes the 90-day program feel like the natural next step. I am building a course-audit practice. Want the full teardown of the email course? Reply and I will send it, no charge.
2 · Chris Reilly · LinkedIn DM only (no email found)
Hi Chris. Instructional designer here; I studied your Mastery Program funnel this week. Your pricing setup is genuinely sharp ($1,244 a la carte against $449 bundled) and the Wharton and Wall Street Prep features add real credibility. Two things you may not see from inside: the About and FAQ pages on the course site 404, and your newsletter site and course site never cross-link, so your newsletter audience has no visible bridge to the $449 offer. I am building a course-audit practice. Want the full read? Happy to send it, no charge.
3 · Jamie Trull · support@jamietrull.com (concierge-gatekept)
Subject: your funnel after the book
Hi Jamie,
I mapped your funnel this week as a working instructional designer. The ladder is genuinely well built: free diagnostics into the book into templates into the Financial Fitness Formula. Hidden Profit as the under-30-dollar anchor is smart, and your blog is current in a niche where most are not.
One structural gap: the Formula opens once a year, so for eleven months your warmest readers have nowhere to go above a $49 template. The old Kajabi store pages are also still live and indexed ("Our Shop Has Moved"), which may split clicks or confuse buyers.
I am building a course-audit practice. Want the full teardown? Reply and I will send it, no charge.
4 · Aron Lange (GRC Lab) · LinkedIn DM ONLY (Germany; UWG/GDPR)
Hi Aron, I went through GRC Lab this week as a working instructional designer. Three things stood out, in a good way. The Lead Implementer curriculum follows a real project arc, twelve steps from management support to audit. The free fundamentals guide actually teaches instead of teasing. And the templates plus the 14-day guarantee make the offer easy to say yes to. One thing may be costing you money right now: when I tried the course platform, learn.grclab.com showed "This school is offline," which would block anyone checking out that way. The homepage also says EUR 997 while the fundamentals page says EUR 347, which makes the price feel unclear. I am building a course-audit practice. Want the full teardown? Reply and I will send it, no charge.
LinkedIn connection note (no-email prospects, 300-char limit)
Hi {first_name}, I am an instructional designer who helps experts pre-sell a course before building it. Saw {specific artifact}. No pitch in this note; happy to share a free read on what a course could look like for your audience if useful.
X intent reply/DM (to someone voicing the problem publicly)
Saw your post about {their words, e.g. whether a course is worth building}. I design courses for a living. The short answer: do not build it until your audience pays for it. Happy to send the 7-step pre-sell checklist I use, free, if you want it.
| Bet | State | Unlocks when |
|---|---|---|
| Partner / affiliate emails (6 audited V3.6 partners, advisory framing; the 50/day affiliate sourcing scrape accumulates silently) | held | first closed client / case study; early partners get 15-25%, white-label scorecard as the integration asset |
| Listen Notes Pro paid podcast sourcing ($200/mo; native email field) — paid tier of agent 11's free sourcing | gated | a 20-email manual test of education-focused podcast hosts produces a booked call or paid Pre-Sell Sprint |
| Apify YouTube About-page email extraction (~$199/mo) | gated | Gate A: first paid Pre-Sell Sprint deposit, or 3+ positive per 300 sends |
| Udemy/Skillshare instructors + post-launch authors + newsletter writers (next ICP tests) | gated | 20-email manual test each, best test wins the next build |
| Founders/agency owners vertical ($1M-10M ARR); clinicians | gated | Day-14 V1 gate per strategy doc; clinicians also need the healthcare-legal review |
| Agencies as paid channel mentors (Hormozi model: learn the channel, bring it in-house, exit) | gated | paid ads era, post offer-proof |
| beehiiv weekly newsletter | gated | landing-page list has subscribers worth migrating |
| Facebook/Messenger outreach | no | not planned; automation gets accounts flagged, no official cold-send API |
Command Center went live 2026-06-11. Layout from a 4-model design debate (Codex, Gemini, DeepSeek, Qwen). KPI bar + cold-email counts are DB-fed; the rest is editorial, edited in this template during reviews.