clembot

the system

An org chart that
has to be true.

Every unit below was read off this repository's own definitions, agent files, command files, the skills registry, the cadence registry. Nothing here is a drawing of what the system might one day do. Where a unit has no receipts, it says so, and most of them have none.

Behind the headline: every seat in the system, placed one at a time. 10 of the 115 stay empty, drawn as footprints with nothing fitted to them, which is the same count the map below reports as vacant.

115units mapped
8with receipts
56defined, not proven
10empty seats
4human gates

The ladder runs the other way

The usual maturity ladder climbs toward autonomy: human-led, then human-assisted, then fully autonomous. This one climbs toward human control, and it tops out at a person saying yes. Nothing here is allowed to be autonomous unless it cannot change anything.

  1. Read-only

    Fetches, classifies, summarises. Changes nothing and sends nothing.

    31 units

  2. Approval-gated

    Changes state or leaves the system. A person approves first.

    15 units

  3. Human-triggered

    Generation. It does not run unless a person runs it.

    44 units

Tier is not a claim we wrote down, it is read off each unit's declared tools. A unit that declares none is assumed gated rather than assumed harmless, because the dangerous error here is overstating autonomy.

The map

Fill is how much autonomy a unit has; stroke is whether it has ever produced a receipt. A tick across a line is a human gate. Select any node to read its card, with scripting off, that jumps to the same card below.

Constellation of 115 units across eight departments. Each node links to its full entry in the list below this diagram.
bip-batch, command, human-triggered, defined not proven/bootstrap-new-project, command, human-triggered, defined not proven/brand-check, command, human-triggered, defined not provenBrand Guardian, agent, approval-gated, defined not proven/campaign, command, human-triggered, defined not provenCampaign Architect, agent, human-triggered, defined not provenCarousel Drafter, agent, human-triggered, defined not provenCode Reviewer, agent, approval-gated, defined not proven/competitor-scan, command, human-triggered, live/content-repurpose, command, human-triggered, defined not proven/content-review, command, human-triggered, defined not provenContent Writer, agent, human-triggered, liveDebugger, agent, approval-gated, defined not proven/deploy, command, human-triggered, defined not proven/design-asset, command, human-triggered, defined not provenDesign Director, agent, human-triggered, live/distribute, command, human-triggered, defined not provendistributor, agent, approval-gated, liveDoc Writer, agent, human-triggered, defined not provendoorman, agent, read-only, defined not proven/dream, command, human-triggered, defined not proven/dream-dry-run, command, human-triggered, defined not provendream-keeper, agent, approval-gated, defined not provenfeed-fetcher, agent, read-only, defined not proven/fix-issue, command, human-triggered, defined not provenfollow-resolver, agent, read-only, defined not provenfollows-digester, agent, read-only, defined not provenideate-leadership, command, human-triggered, defined not proven/launch-network, command, human-triggered, defined not provenleadership-persona, agent, approval-gated, defined not provenLinkedIn Drafter, agent, human-triggered, liveMarket Analyst, agent, human-triggered, liveMarketing Dispatcher, agent, human-triggered, livemid-session-save, command, human-triggered, defined not provenClembot-Brain, persona, read-only, liveClembot-Dev, persona, read-only, liveClembot-Marketing, persona, read-only, liveClembot-Publisher, persona, read-only, liveClembot-Research, persona, read-only, liveClembot-Sales, persona, read-only, livepost-summarizer, agent, read-only, defined not proven/pr-review, command, human-triggered, defined not proven/project, command, human-triggered, defined not provenpublish-thought, command, human-triggered, defined not provenPublisher, agent, approval-gated, defined not proven/recall-search <query>, command, human-triggered, defined not provenRefactorer, agent, approval-gated, defined not provenAmbient conversations, routine, read-only, liveAnatomy case-study ingest, routine, human-triggered, live/bip-batch, routine, human-triggered, liveBuild brief export, routine, read-only, liveContent-generation commands, routine, human-triggered, livecontent sync-postiz, routine, read-only, liveDaily Product Pulse, routine, read-only, liveFollow / Podcast poller, routine, read-only, liveGmail Poller, routine, read-only, liveGmail Scout, routine, read-only, liveList30 Content Loop, routine, approval-gated, liveNewsroom standup, routine, approval-gated, liveNotion Calendar Watcher, routine, read-only, livePOW Debt Scanner, routine, read-only, livePublic site feed export, routine, read-only, livePublic site map export, routine, read-only, liveSabbatical Weekly Digest, routine, read-only, liveStack Supervisor, routine, read-only, liveStale Deploy Detector, routine, read-only, liveStale Draft Checker, routine, read-only, liveSTATE.md Daily Sync, routine, read-only, liveUpcoming Drafts Digest, routine, read-only, liveWeekly LinkedIn build log, routine, approval-gated, liveWeekly LinkedIn Draft — RETIRED 2026-07-26, routine, read-only, supersededWeekly metrics pull, routine, read-only, liveSabbatical Tracker, agent, human-triggered, defined not provenSecurity Auditor, agent, approval-gated, defined not proven/seo-pulse, command, human-triggered, defined not provenSEO Specialist, agent, human-triggered, defined not proven/session-end, command, human-triggered, defined not proven/setup-client-routines, command, human-triggered, defined not provenshare-poster, agent, approval-gated, defined not provenarticle-writing, skill, not executable, livecampaign, skill, not executable, defined not provencompetitor-scan, skill, not executable, defined not provencontent-engine, skill, not executable, livecontinuous-learning, skill, not executable, livelaunch-network, skill, not executable, livemarket-research, skill, not executable, livemarketingskills, skill, not executable, liverebrand-asset, skill, not executable, defined not provenseo, skill, not executable, liveseo-pulse, skill, not executable, defined not provensuperpowers, skill, not executable, livetweet-thread, skill, not executable, liveweekly-report, skill, not executable, defined not provenwrite-blog, skill, not executable, liveSocial Strategist, agent, human-triggered, live/status, command, human-triggered, defined not proven/sync-state, command, human-triggered, defined not provenTest Writer, agent, approval-gated, defined not proventhought-leadership, command, human-triggered, defined not proventhoughts-publisher, agent, approval-gated, defined not proven/tweet-thread, command, human-triggered, defined not provenurl-unfurler, agent, read-only, defined not provenAsset librarian, vacancy, not executable, vacantIncident response, vacancy, not executable, vacantCost accounting, vacancy, not executable, vacantInvoicing, vacancy, not executable, vacantRoadmap / spec, vacancy, not executable, vacantUser research, vacancy, not executable, vacantCRM sync, vacancy, not executable, vacantLead qualification, vacancy, not executable, vacantOutreach sequencer, vacancy, not executable, vacantProposal / SOW writer, vacancy, not executable, vacant/vet, command, human-triggered, defined not proven/weekly-report, command, human-triggered, defined not proven/write-blog, command, human-triggered, defined not proven

The roster

C-suite18 staffed · 0 vacant

/dream

command · C-suite · human-triggered

NOT YET PROVEN

Run a memory consolidation pass over Claude Code memory and the Obsidian ClemVault second brain

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/commands/dream.md on 2026-09-21

/dream-dry-run

command · C-suite · human-triggered

NOT YET PROVEN

Preview what /dream would change without writing any files.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/commands/dream-dry-run.md on 2026-09-21

dream-keeper

agent · C-suite · approval-gated

NOT YET PROVEN

Subagent dedicated to memory consolidation passes.

The human

A person approves before anything leaves the system.

  1. Read-only
  2. Approval-gated← here
  3. Human-triggered

Tier not derivable, this unit declares no tools, so it is assumed gated rather than assumed harmless.

What it replaces

Notes that accumulate until nobody can find anything in them.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/agents/dream-keeper.md on 2026-09-21

Marketing Dispatcher

agent · C-suite · human-triggered

Use this agent as the orchestrator when a high-level marketing command needs to be broken into a multi-agent workflow.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here
Breaks into
Brand GuardianCampaign ArchitectContent WriterDesign DirectorMarket AnalystPublisherSEO SpecialistSocial Strategist

Embodies: log is the product

What it replaces

Deciding, per task, which of eleven things to do first.

Receipts

2026-05-16weekly-reportRunning
2026-05-16write-blogRunning
2026-05-06campaignRunning

3 of 10 shown · what these prove

Plus 1 record from a seeded demonstration, written in one batch against sample data. Not counted as receipts.

generated from .claude/agents/marketing-dispatcher.md on 2026-09-21

mid-session-save

command · C-suite · human-triggered

NOT YET PROVEN

Lightweight mid-session checkpoint — saves state to primer.md, memory/, and tasks/todo.md without committing or deploying.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/commands/mid-session-save.md on 2026-09-21

Clembot-Brain

persona · C-suite · read-only

strategy and synthesis — how this fits the positioning, what decision it implies, what the second-order effect is.

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from scripts/lib/personas.js on 2026-09-21

/project

command · C-suite · human-triggered

NOT YET PROVEN

Load a project from the vault.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/commands/project.md on 2026-09-21

Ambient conversations

routine · C-suite · read-only

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from docs/automation-registry.md#cadence-registry on 2026-09-21

Build brief export

routine · C-suite · read-only

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from docs/automation-registry.md#cadence-registry on 2026-09-21

Sabbatical Weekly Digest

routine · C-suite · read-only

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from docs/automation-registry.md#cadence-registry on 2026-09-21

STATE.md Daily Sync

routine · C-suite · read-only

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Embodies: code and data apart

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from docs/automation-registry.md#cadence-registry on 2026-09-21

Sabbatical Tracker

agent · C-suite · human-triggered

NOT YET PROVEN

Use this agent to write weekly sabbatical log entries, update the milestone tracker, and surface the key progress moment for LinkedIn.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/agents/sabbatical-tracker.md on 2026-09-21

/session-end

command · C-suite · human-triggered

NOT YET PROVEN

End the current session cleanly.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/commands/session-end.md on 2026-09-21

continuous-learning

skill · C-suite

Community pack — UNVETTED, source unknown

The human

A person approves before anything leaves the system.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from skills/continuous-learning/SKILL.md on 2026-09-21

weekly-report

skill · C-suite

NOT YET PROVEN

Weekly report pack (pairs with /weekly-report)

The human

A person approves before anything leaves the system.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/rules/security.md on 2026-09-21

/status

command · C-suite · human-triggered

NOT YET PROVEN

Print a compact project health check.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/commands/status.md on 2026-09-21

/sync-state

command · C-suite · human-triggered

NOT YET PROVEN

Regenerate STATE.md from current repo reality.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/commands/sync-state.md on 2026-09-21

/weekly-report

command · C-suite · human-triggered

NOT YET PROVEN

Generate the weekly marketing performance and intelligence summary.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here
Breaks into
Market AnalystSEO Specialist

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/commands/weekly-report.md on 2026-09-21

Marketing29 staffed · 0 vacant

bip-batch

command · Marketing · human-triggered

NOT YET PROVEN

Draft one week of build-in-public content (Mon/Wed/Fri) for the clembot tenant — LinkedIn + X — as 3 Slack approval cards.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here
Breaks into
Brand GuardianLinkedIn DrafterSocial Strategist

Embodies: human gate

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/commands/bip-batch.md on 2026-09-21

/campaign

command · Marketing · human-triggered

NOT YET PROVEN

Plan and scaffold a complete multi-channel marketing campaign with content calendar, KPIs, and per-piece briefs.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here
Breaks into
Market Analyst

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/commands/campaign.md on 2026-09-21

Campaign Architect

agent · Marketing · human-triggered

NOT YET PROVEN

Use this agent to plan multi-channel marketing campaigns, build content calendars, define KPIs, and sequence content across the funnel.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here
Breaks into
Content WriterPublisher
Gated by
Brand Guardian
Called by
Market AnalystMarketing DispatcherSEO Specialist

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/agents/campaign-architect.md on 2026-09-21

/content-repurpose

command · Marketing · human-triggered

NOT YET PROVEN

Take an existing piece of content and repurpose it into multiple channel-optimized formats.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here
Breaks into
Content WriterSocial Strategist

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/commands/content-repurpose.md on 2026-09-21

/content-review

command · Marketing · human-triggered

NOT YET PROVEN

Ingest content performance data and feed it back into the wiki.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here
Breaks into
Publisher

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/commands/content-review.md on 2026-09-21

Content Writer

agent · Marketing · human-triggered

Use this agent to produce long-form content — blog posts, case studies, docs, email copy, and product newsletters.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here
Gated by
Brand Guardian
Called by
/bootstrap-new-projectCampaign Architect/content-repurpose/launch-networkMarketing DispatcherSEO Specialist

What it replaces

A blank page on a Sunday night.

Receipts

2026-03-19/campaign-newsletterRunning
2026-03-19/campaign-newsletterComplete
2026-03-19/campaign-newsletterRunning

3 of 5 shown · 1 withheld (they carry a client or platform id) · what these prove

Plus 2 records from a seeded demonstration, written in one batch against sample data. Not counted as receipts.

generated from .claude/agents/content-writer.md on 2026-09-21

ideate-leadership

command · Marketing · human-triggered

NOT YET PROVEN

Scans market trends and identifies high-signal topics for founder thought leadership.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/commands/ideate-leadership.md on 2026-09-21

/launch-network

command · Marketing · human-triggered

NOT YET PROVEN

Generate a complete network launch package: blog post, Twitter thread, LinkedIn post, and network card visual — all in one command.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here
Breaks into
Brand GuardianContent WriterDesign DirectorPublisherSocial Strategist

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/commands/launch-network.md on 2026-09-21

leadership-persona

agent · Marketing · approval-gated

NOT YET PROVEN

The human

A person approves before anything leaves the system.

  1. Read-only
  2. Approval-gated← here
  3. Human-triggered

Tier not derivable, this unit declares no tools, so it is assumed gated rather than assumed harmless.

What it replaces

Writing the thought-leadership post yourself, in a voice you borrowed from LinkedIn.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/agents/leadership-persona.md on 2026-09-21

LinkedIn Drafter

agent · Marketing · human-triggered

Use this agent to draft LinkedIn posts in Clemens Wan's voice.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here
Called by
bip-batch

What it replaces

Writing in a voice that drifts a little further from your own every week.

Receipts

2026-09-20write-linkedin-postComplete
2026-09-13write-linkedin-postComplete
2026-09-06write-linkedin-postComplete

3 of 17 shown · what these prove

generated from .claude/agents/linkedin-drafter.md on 2026-09-21

Clembot-Marketing

persona · Marketing · read-only

content angles, hooks, distribution, audience reach.

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from scripts/lib/personas.js on 2026-09-21

/bip-batch

routine · Marketing · human-triggered

The human

Human-run. Produces 3 drafts and 3 Slack approval cards — nothing schedules itself.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from docs/automation-registry.md#cadence-registry on 2026-09-21

Content-generation commands

routine · Marketing · human-triggered

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from docs/automation-registry.md#cadence-registry on 2026-09-21

List30 Content Loop

routine · Marketing · approval-gated

The human

Generates on a timer into DRAFT only. A human approves every item before Postiz.

  1. Read-only
  2. Approval-gated← here
  3. Human-triggered

Embodies: human gate

What it replaces

Three evenings a week making carousels by hand.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from docs/automation-registry.md#cadence-registry on 2026-09-21

Weekly LinkedIn build log

routine · Marketing · approval-gated

The human

A person approves before anything leaves the system.

  1. Read-only
  2. Approval-gated← here
  3. Human-triggered

What it replaces

Trying to remember on Sunday what you shipped on Tuesday.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from docs/automation-registry.md#cadence-registry on 2026-09-21

Weekly LinkedIn Draft — RETIRED 2026-07-26

routine · Marketing · read-only

SUPERSEDED

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

What it replaces

A Sunday-night draft nobody asked for. Replaced by a batch that at least waits to be asked.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from docs/automation-registry.md#cadence-registry on 2026-09-21

Weekly metrics pull

routine · Marketing · read-only

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from docs/automation-registry.md#cadence-registry on 2026-09-21

article-writing

skill · Marketing

Long-form article patterns

The human

A person approves before anything leaves the system.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from skills/article-writing/SKILL.md on 2026-09-21

campaign

skill · Marketing

NOT YET PROVEN

Campaign planning pack (pairs with /campaign)

The human

A person approves before anything leaves the system.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/rules/security.md on 2026-09-21

content-engine

skill · Marketing

Content pipeline pack

The human

A person approves before anything leaves the system.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from skills/content-engine/SKILL.md on 2026-09-21

launch-network

skill · Marketing

Launch package workflow

The human

A person approves before anything leaves the system.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from skills/launch-network/SKILL.md on 2026-09-21

marketingskills

skill · Marketing

Community bundle — UNVETTED, source unknown

The human

A person approves before anything leaves the system.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from skills/marketingskills/SKILL.md on 2026-09-21

tweet-thread

skill · Marketing

Thread drafting

The human

A person approves before anything leaves the system.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from skills/tweet-thread/SKILL.md on 2026-09-21

write-blog

skill · Marketing

Blog drafting

The human

A person approves before anything leaves the system.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from skills/write-blog/SKILL.md on 2026-09-21

Social Strategist

agent · Marketing · human-triggered

Use this agent to write short-form content — Twitter/X threads, LinkedIn posts, EG founder persona content, and community replies.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here
Gated by
Brand Guardian
Called by
bip-batch/content-repurpose/launch-networkMarketing Dispatcher/tweet-thread

What it replaces

Guessing at a hook and posting it anyway.

Receipts

2026-03-26/competitor-scanComplete
2026-03-19/campaign-newsletterComplete
2026-03-19/campaign-newsletterComplete

3 of 5 shown · what these prove

generated from .claude/agents/social-strategist.md on 2026-09-21

thought-leadership

command · Marketing · human-triggered

NOT YET PROVEN

Converts a high-signal trend into a multi-channel thought leadership draft (LinkedIn + X).

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/commands/thought-leadership.md on 2026-09-21

/tweet-thread

command · Marketing · human-triggered

NOT YET PROVEN

Write a Twitter/X thread on the given topic with a high-performing hook.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here
Breaks into
Social Strategist

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/commands/tweet-thread.md on 2026-09-21

/write-blog

command · Marketing · human-triggered

NOT YET PROVEN

Produce a fully SEO-optimized, brand-compliant blog post on the given topic.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here
Breaks into
SEO Specialist

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/commands/write-blog.md on 2026-09-21

Sales1 staffed · 4 vacant

Clembot-Sales

persona · Sales · read-only

pipeline, conversion, client objections, demos.

The human

Comments in threads. Has no ability to contact anyone.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from scripts/lib/personas.js on 2026-09-21

CRM sync

vacancy · Sales

EMPTY SEAT

The databases exist and are mis-pointed at each other; no agent owns them.

The human

Nobody. That is what this entry is recording.

Receipts

None, nothing has ever run here.

generated from site/src/data/map-overlay.yaml on 2026-09-21

Lead qualification

vacancy · Sales

EMPTY SEAT

No unit matches. Nothing in the repo scores, sorts or routes an inbound lead.

The human

Nobody. That is what this entry is recording.

Receipts

None, nothing has ever run here.

generated from site/src/data/map-overlay.yaml on 2026-09-21

Outreach sequencer

vacancy · Sales

EMPTY SEAT

No unit matches — deliberately. Nothing here is allowed to contact a stranger.

The human

Nobody. That is what this entry is recording.

Receipts

None, nothing has ever run here.

generated from site/src/data/map-overlay.yaml on 2026-09-21

Proposal / SOW writer

vacancy · Sales

EMPTY SEAT

No unit matches. Proposals are written by hand, outside the system.

The human

Nobody. That is what this entry is recording.

Receipts

None, nothing has ever run here.

generated from site/src/data/map-overlay.yaml on 2026-09-21

Product & ops15 staffed · 2 vacant

/bootstrap-new-project

command · Product & ops · human-triggered

NOT YET PROVEN

Scaffold a complete marketing operating system for a new project from the intake questionnaire.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here
Breaks into
Brand GuardianContent WriterPublisher

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/commands/bootstrap-new-project.md on 2026-09-21

/distribute

command · Product & ops · human-triggered

NOT YET PROVEN

Run a distribution pass for one published post.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here

Embodies: human gate

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/commands/distribute.md on 2026-09-21

distributor

agent · Product & ops · approval-gated

Formats a published post for a target distribution channel.

The human

Formats for a channel. Never posts — a human releases it.

  1. Read-only
  2. Approval-gated← here
  3. Human-triggered

Embodies: human gate

What it replaces

Rewriting the same post by hand for every channel, at the end of a day, badly.

Receipts

2026-06-26distributeComplete
2026-06-25distributeComplete
2026-06-25distributeComplete

3 of 61 shown · 55 withheld (they carry a client or platform id) · what these prove

generated from .claude/agents/distributor.md on 2026-09-21

Clembot-Publisher

persona · Product & ops · read-only

cadence, scheduling, the pipeline from draft to posted.

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from scripts/lib/personas.js on 2026-09-21

publish-thought

command · Product & ops · human-triggered

NOT YET PROVEN

1.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here

Embodies: human gate

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/commands/publish-thought.md on 2026-09-21

Publisher

agent · Product & ops · approval-gated

NOT YET PROVEN

Final distribution agent — saves approved content to outputs/, syncs to Notion, optionally publishes to Ghost CMS, and archives the full evidence chain.

The human

Only publishes what brand-guardian already approved.

  1. Read-only
  2. Approval-gated← here
  3. Human-triggered
Called by
/bootstrap-new-projectBrand GuardianCampaign Architect/content-review/launch-networkMarketing Dispatcher

Embodies: human gateone writer

What it replaces

Copy-pasting an approved draft into four places, then forgetting which one was the record.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

Plus 2 records from a seeded demonstration, written in one batch against sample data. Not counted as receipts.

generated from .claude/agents/publisher.md on 2026-09-21

Anatomy case-study ingest

routine · Product & ops · human-triggered

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here

Embodies: allowlist fail closed

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from docs/automation-registry.md#cadence-registry on 2026-09-21

content sync-postiz

routine · Product & ops · read-only

The human

Reconciles state after the fact. Never initiates a post.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Embodies: one writer

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from docs/automation-registry.md#cadence-registry on 2026-09-21

Notion Calendar Watcher

routine · Product & ops · read-only

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Embodies: one writer

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from docs/automation-registry.md#cadence-registry on 2026-09-21

Public site feed export

routine · Product & ops · read-only

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Embodies: allowlist fail closedcode and data apartlog is the product

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from docs/automation-registry.md#cadence-registry on 2026-09-21

Public site map export

routine · Product & ops · read-only

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Embodies: allowlist fail closedlog is the product

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from docs/automation-registry.md#cadence-registry on 2026-09-21

Stale Draft Checker

routine · Product & ops · read-only

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from docs/automation-registry.md#cadence-registry on 2026-09-21

Upcoming Drafts Digest

routine · Product & ops · read-only

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from docs/automation-registry.md#cadence-registry on 2026-09-21

/setup-client-routines

command · Product & ops · human-triggered

NOT YET PROVEN

Set up CCR routines for a client.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here

Embodies: code and data apart

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/commands/setup-client-routines.md on 2026-09-21

thoughts-publisher

agent · Product & ops · approval-gated

NOT YET PROVEN

The human

Runs only on a Thought a human already moved to Approved.

  1. Read-only
  2. Approval-gated← here
  3. Human-triggered

Embodies: human gate

What it replaces

A manual commit, a manual deploy, and a page nobody remembered to mark Published.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/agents/thoughts-publisher.md on 2026-09-21

Roadmap / spec

vacancy · Product & ops

EMPTY SEAT

Ops is staffed; product is not. Scope lives in a human head and a todo file.

The human

Nobody. That is what this entry is recording.

Receipts

None, nothing has ever run here.

generated from site/src/data/map-overlay.yaml on 2026-09-21

User research

vacancy · Product & ops

EMPTY SEAT

No unit talks to a user. Every signal here is second-hand.

The human

Nobody. That is what this entry is recording.

Receipts

None, nothing has ever run here.

generated from site/src/data/map-overlay.yaml on 2026-09-21

Development17 staffed · 1 vacant

Code Reviewer

agent · Development · approval-gated

NOT YET PROVEN

Use this agent to review code changes for correctness, maintainability, performance, and adherence to project conventions before merging.

The human

A person approves before anything leaves the system.

  1. Read-only
  2. Approval-gated← here
  3. Human-triggered
Breaks into
Security AuditorTest Writer
Called by
Debugger/pr-review

What it replaces

Merging your own work and finding out later.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/agents/code-reviewer.md on 2026-09-21

Debugger

agent · Development · approval-gated

NOT YET PROVEN

Use this agent to diagnose runtime errors, trace component state issues, debug API failures, and identify the root cause of unexpected behaviour.

The human

A person approves before anything leaves the system.

  1. Read-only
  2. Approval-gated← here
  3. Human-triggered
Breaks into
Code Reviewer
Called by
/fix-issue

What it replaces

A console.log, moved around until the problem moves.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/agents/debugger.md on 2026-09-21

/deploy

command · Development · human-triggered

NOT YET PROVEN

Build the dashboard-app for production and optionally run a preview server or deploy to a hosting target.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/commands/deploy.md on 2026-09-21

Doc Writer

agent · Development · human-triggered

NOT YET PROVEN

Use this agent to generate or update documentation — README sections, JSDoc comments, setup guides, API docs, and inline code documentation.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/agents/doc-writer.md on 2026-09-21

doorman

agent · Development · read-only

NOT YET PROVEN

Grades an unknown MCP server before you let it into your session.

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Embodies: tool text is untrustedhuman gate

What it replaces

Reading a tool's README and hoping.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/agents/doorman.md on 2026-09-21

/fix-issue

command · Development · human-triggered

NOT YET PROVEN

Structured workflow for fixing a bug or issue: reproduce, diagnose, fix, verify, document.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here
Breaks into
Debugger

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/commands/fix-issue.md on 2026-09-21

Clembot-Dev

persona · Development · read-only

feasibility, automation leverage, what to build or instrument.

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from scripts/lib/personas.js on 2026-09-21

/pr-review

command · Development · human-triggered

NOT YET PROVEN

Run a structured code review on the current branch diff or a specific PR.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here
Breaks into
Code Reviewer

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/commands/pr-review.md on 2026-09-21

Refactorer

agent · Development · approval-gated

NOT YET PROVEN

Use this agent to restructure code while preserving behaviour — extract components, deduplicate logic, improve type safety, simplify state management, and clean up technical debt.

The human

A person approves before anything leaves the system.

  1. Read-only
  2. Approval-gated← here
  3. Human-triggered

What it replaces

Meaning to come back to it.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/agents/refactorer.md on 2026-09-21

Daily Product Pulse

routine · Development · read-only

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from docs/automation-registry.md#cadence-registry on 2026-09-21

POW Debt Scanner

routine · Development · read-only

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from docs/automation-registry.md#cadence-registry on 2026-09-21

Stack Supervisor

routine · Development · read-only

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from docs/automation-registry.md#cadence-registry on 2026-09-21

Stale Deploy Detector

routine · Development · read-only

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from docs/automation-registry.md#cadence-registry on 2026-09-21

Security Auditor

agent · Development · approval-gated

NOT YET PROVEN

Use this agent to audit code for security vulnerabilities — exposed secrets, XSS vectors, unsafe API patterns, dependency risks, and OWASP top 10 issues.

The human

A person approves before anything leaves the system.

  1. Read-only
  2. Approval-gated← here
  3. Human-triggered
Called by
Code Reviewer

Embodies: allowlist fail closedverify not guess

What it replaces

Assuming the secrets are still where you left them.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/agents/security-auditor.md on 2026-09-21

superpowers

skill · Development

Community bundle (obra/superpowers) — UNVETTED

The human

A person approves before anything leaves the system.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from skills/superpowers/SKILL.md on 2026-09-21

Test Writer

agent · Development · approval-gated

NOT YET PROVEN

Use this agent to write unit tests, integration tests, and component tests for the dashboard-app (React/TypeScript) and Node.js scripts.

The human

A person approves before anything leaves the system.

  1. Read-only
  2. Approval-gated← here
  3. Human-triggered
Called by
Code Reviewer

What it replaces

Testing it by using it.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/agents/test-writer.md on 2026-09-21

Incident response

vacancy · Development

EMPTY SEAT

A supervisor restarts what is down. Nothing diagnoses why, or tells anyone.

The human

Nobody. That is what this entry is recording.

Receipts

None, nothing has ever run here.

generated from site/src/data/map-overlay.yaml on 2026-09-21

/vet

command · Development · human-triggered

NOT YET PROVEN

Vet a candidate before an agent gets it.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here

Embodies: tool text is untrustedverify not guess

What it replaces

Adding a server because it was the first search result.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/commands/vet.md on 2026-09-21

Brand5 staffed · 1 vacant

/brand-check

command · Brand · human-triggered

NOT YET PROVEN

Run a full brand compliance audit on any piece of content.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/commands/brand-check.md on 2026-09-21

Brand Guardian

agent · Brand · approval-gated

NOT YET PROVEN

Use this agent to review any content for brand compliance, voice alignment, positioning accuracy, and factual correctness before publishing.

The human

Nothing reaches a public channel without a PASS. Veto power, no override.

  1. Read-only
  2. Approval-gated← here
  3. Human-triggered
Breaks into
Publisher
Called by
bip-batch/bootstrap-new-project/launch-networkMarketing Dispatcher
Gates
Campaign ArchitectContent WriterDesign DirectorSocial Strategist

Embodies: human gateverify not guess

What it replaces

A final read-through by whoever was free, and the hope that nobody shipped a claim we could not back.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/agents/brand-guardian.md on 2026-09-21

/design-asset

command · Brand · human-triggered

NOT YET PROVEN

Generate a brand-compliant visual asset using Canva.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here
Breaks into
Design Director

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/commands/design-asset.md on 2026-09-21

Design Director

agent · Brand · human-triggered

Use this agent to create visual assets — network cards, social graphics, landing page sections, and campaign visuals.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here
Gated by
Brand Guardian
Called by
/design-asset/launch-networkMarketing Dispatcher

What it replaces

A design session that starts from an empty template every time.

Receipts

2026-03-19/campaign-newsletterComplete
2026-03-19/campaign-newsletterComplete

2 of 2 shown · what these prove

generated from .claude/agents/design-director.md on 2026-09-21

rebrand-asset

skill · Brand

NOT YET PROVEN

Rebrand asset pack

The human

A person approves before anything leaves the system.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/rules/security.md on 2026-09-21

Asset librarian

vacancy · Brand

EMPTY SEAT

Assets are generated, then lost. Nothing catalogues or reuses them.

The human

Nobody. That is what this entry is recording.

Receipts

None, nothing has ever run here.

generated from site/src/data/map-overlay.yaml on 2026-09-21

Research20 staffed · 0 vacant

/competitor-scan

command · Research · human-triggered

Deep competitive intelligence on named competitors, producing counter-positioning recommendations.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here
Breaks into
Market Analyst

Embodies: verify not guess

What it replaces

Checking a competitor when you happen to remember they exist.

Receipts

2026-03-26/competitor-scanComplete

1 of 1 shown · what these prove

generated from .claude/commands/competitor-scan.md on 2026-09-21

feed-fetcher

agent · Research · read-only

NOT YET PROVEN

Pure RSS/Atom fetch and parse.

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/agents/feed-fetcher.md on 2026-09-21

follow-resolver

agent · Research · read-only

NOT YET PROVEN

Converts a raw user input (URL, domain, or handle) into a validated RSS/Atom

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/agents/follow-resolver.md on 2026-09-21

follows-digester

agent · Research · read-only

NOT YET PROVEN

Takes N summarized posts from one feed and formats them as a Slack digest —

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/agents/follows-digester.md on 2026-09-21

Market Analyst

agent · Research · human-triggered

Gathers competitive intelligence, network research, market sizing, and SEO gaps — producing research briefs and competitor scan reports to inform the content team.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here
Breaks into
Campaign Architect
Called by
/campaign/competitor-scanMarketing Dispatcher/weekly-report

Embodies: verify not guess

What it replaces

Twelve open tabs and no notes.

Receipts

2026-05-06/competitor-scanRunning
2026-05-06/competitor-scanComplete
2026-03-26/competitor-scanRunning

3 of 7 shown · what these prove

Plus 13 records from a seeded demonstration, written in one batch against sample data. Not counted as receipts.

generated from .claude/agents/market-analyst.md on 2026-09-21

Clembot-Research

persona · Research · read-only

evidence, market patterns, competitors, precedent.

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from scripts/lib/personas.js on 2026-09-21

post-summarizer

agent · Research · read-only

NOT YET PROVEN

Produces a 2-3 sentence summary and a one-sentence key insight from a

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/agents/post-summarizer.md on 2026-09-21

Follow / Podcast poller

routine · Research · read-only

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from docs/automation-registry.md#cadence-registry on 2026-09-21

Gmail Poller

routine · Research · read-only

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from docs/automation-registry.md#cadence-registry on 2026-09-21

Gmail Scout

routine · Research · read-only

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from docs/automation-registry.md#cadence-registry on 2026-09-21

Newsroom standup

routine · Research · approval-gated

The human

A person approves before anything leaves the system.

  1. Read-only
  2. Approval-gated← here
  3. Human-triggered

What it replaces

Reading nine feeds and forgetting what was in the first one.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from docs/automation-registry.md#cadence-registry on 2026-09-21

/seo-pulse

command · Research · human-triggered

NOT YET PROVEN

Scan trending topics, surface rising keywords, and produce a prioritized list of content recommendations for the week.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here
Breaks into
SEO Specialist

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/commands/seo-pulse.md on 2026-09-21

SEO Specialist

agent · Research · human-triggered

NOT YET PROVEN

Use this agent for SEO research, keyword analysis, content gap identification, SERP analysis, and search-optimized content recommendations.

The human

It does not run unless a person runs it.

  1. Read-only
  2. Approval-gated
  3. Human-triggered← here
Breaks into
Campaign ArchitectContent Writer
Called by
Marketing Dispatcher/seo-pulse/weekly-report/write-blog

Embodies: verify not guess

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/agents/seo-specialist.md on 2026-09-21

share-poster

agent · Research · approval-gated

NOT YET PROVEN

Formats a single unfurled + summarized URL as a Slack message for #clembot-shares.

The human

Posts only what a human explicitly shared.

  1. Read-only
  2. Approval-gated← here
  3. Human-triggered

Embodies: human gate

What it replaces

A link dropped in a channel and never opened again.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/agents/share-poster.md on 2026-09-21

competitor-scan

skill · Research

NOT YET PROVEN

Competitive scan pack (pairs with /competitor-scan)

The human

A person approves before anything leaves the system.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/rules/security.md on 2026-09-21

market-research

skill · Research

Research workflow pack

The human

A person approves before anything leaves the system.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from skills/market-research/SKILL.md on 2026-09-21

seo

skill · Research

SEO analysis pack

The human

A person approves before anything leaves the system.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from skills/seo/SKILL.md on 2026-09-21

seo-pulse

skill · Research

NOT YET PROVEN

Weekly SEO scan pack (pairs with /seo-pulse)

The human

A person approves before anything leaves the system.

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/rules/security.md on 2026-09-21

url-unfurler

agent · Research · read-only

NOT YET PROVEN

Fetches any URL and extracts structured metadata — title, excerpt, og:image,

The human

No gate needed, this unit cannot change anything or send anything.

  1. Read-only← here
  2. Approval-gated
  3. Human-triggered

Receipts

Defined, not yet proven. Nothing in the evidence chain has run under this name.

generated from .claude/agents/url-unfurler.md on 2026-09-21

Finance0 staffed · 2 vacant

Cost accounting

vacancy · Finance

EMPTY SEAT

Token spend and money were never instrumented. The Proof page says so too.

The human

Nobody. That is what this entry is recording.

Receipts

None, nothing has ever run here.

generated from site/src/data/map-overlay.yaml on 2026-09-21

Invoicing

vacancy · Finance

EMPTY SEAT

Invoices live as files in the vault. No unit reads or produces them.

The human

Nobody. That is what this entry is recording.

Receipts

None, nothing has ever run here.

generated from site/src/data/map-overlay.yaml on 2026-09-21