Standalone article · part of a sequenced guide
What you'll unlock: Serious Cowork users do not brainstorm workflows from scratch — they clone blueprints, map paths, and promote through production gates.
End-to-End Workflow Blueprints
Complete, detailed, production-ready workflow blueprints — the reference library for serious Cowork users
Chapter context
You know how to build Skills — but staring at an empty workflow builder wastes the hour you were trying to save. Chapter 6 gives production-ready designs
Is this chapter for you?
Are you automating leadership, ops, or content work?
Start in Concept 1, 2, or 3 — clone closest blueprint before custom design.
Does the blueprint touch external sends or money?
Default T3 approval from blueprint; read Ch 5 §2.4 before enabling send nodes.
Do you already have a similar workflow from Ch 3–4?
Merge duplicate Skills — blueprints share DOC_PIPELINE, KPI_REPORT, intel scans.
Is this your first production workflow?
Start §2.1 document pipeline or §1.1 exec brief (read-only connectors) — highest learning ROI.
Chapters 1–5 taught mental models, Skills, schedules, connectors, and production discipline. Chapter 6 is the blueprint library — executive, operations, and content/knowledge patterns you implement rather than invent.Each blueprint references prior chapters: TAR Skills (Ch 2), triggers (Ch 3), MCP (Ch 4), validation and approval gates (Ch 5). Clone, allowlist, regression-test, then schedule.
Chapter insight
Serious Cowork users do not brainstorm workflows from scratch — they clone blueprints, map paths, and promote through production gates.
Reference diagrams
Blueprint library map
Three lanes — pick blueprint by role, adapt paths and connectors.
Blueprint adaptation loop
Clone spec → map paths → sandbox → production gate → schedule.
Implementation paths
Reference designs — not generic inspiration slides.
Concept 1
Executive & Leadership Workflows
The operational workflows that give senior leaders back hours every week — complete designs with Skill specifications
1.1
The executive morning brief
Calendar + email + news + Slack → prioritised daily briefing delivered by 8am
Key takeaway
Blueprint EXEC_BRIEF_v1: four parallel fetches → JOIN → prioritise → one template — delivered to Drive + Slack link by 7:45am, HITL T2.
Why this matters
This is the flagship leadership workflow — proves MCP + pipeline discipline from Chapters 3–5 in one package.
Pipeline: CAL_FETCH || GMAIL_FETCH || SLACK_READ || NEWS_SCAN → JOIN → PRIORITISE_v1 → BRIEF_EXEC_v1 → VALIDATE → DRIVE_WRITE + SLACK_NOTIFY.
Schedule: cron 6:30 local. Output sections: Top 3 decisions, Calendar risks, Overnight signals, Waiting on you. Token budget: cap fetches; degrade per Ch 5 §1.4.
Workflow — do this next
- 01Clone blueprint spec; set connector allowlists.
- 02Run manual ×5 with production readiness gate.
- 03Schedule 6:30; notify #exec-brief with Drive link.
- 04Executive reviews by 8am — never auto-forward externally.
Ready-to-use artifacts
Complete templates — paste directly into your AI tool or automation workflow.
EXEC_MORNING_BRIEF blueprint
Skills: CAL_FETCH | GMAIL_FETCH | SLACK_READ | NEWS_SCAN
→ JOIN → PRIORITISE_v1 → BRIEF_EXEC_v1 → VALIDATE → EMIT
Schedule: 06:30 | HITL: T2 | SLA: brief by 07:45
Output: ~/Briefs/exec/{date}/brief.md1.2
The board pack preparation workflow
Multiple input documents → synthesised executive summary → formatted presentation → distributed
Key takeaway
BOARD_PACK_v1: ingest fixed folder → extract metrics per doc type → SYNTH_BOARD → FORMAT_DECK outline — staging until T3 approval before distribution.
Why this matters
Board week is the highest-stakes ops window — this blueprint encodes change freeze and validation from Ch 5.
Trigger: condition board_inputs_ready.flag in ~/Board/inbox/{quarter}/. Skills: INGEST_BOARD_DOCS → EXTRACT_FINANCE | EXTRACT_OPS | EXTRACT_PRODUCT → SYNTH_BOARD_v1 → FORMAT_BOARD_MEMO_v1 + slide_outline.md. QA reconciles financial totals to source.
Non-goals: never auto-send to board portal. Distribution only after approval.flag + partner review. Archive prior quarter for regression fixtures.
Workflow — do this next
- 01Define board folder schema + flag lifecycle.
- 02Build extract Skills per document type.
- 03QA gate on financial reconciliation.
- 04T3 approval before any connector write to board share.
Ready-to-use artifacts
Complete templates — paste directly into your AI tool or automation workflow.
BOARD_PACK pipeline spec
INGEST → EXTRACT_* → SYNTH_BOARD → QA → FORMAT → REVIEW_QUEUE
→ (T3) DRIVE_DISTRIBUTE
Freeze: no Skill version bumps 7d before board1.3
The stakeholder communication workflow
Update inputs → personalised stakeholder messages → review queue → approved sends
Key takeaway
STAKEHOLDER_COMMS_v1: segment map in Context → one draft per stakeholder tier → queue — send only with per-recipient approval.flag.
Why this matters
Personalised at scale without uncontrolled send is the leadership comms sweet spot.
Inputs: weekly update bullets, CRM segment export, STAKEHOLDER_MAP.json (tier, tone, channel). CLASSIFY_STAKEHOLDER → DRAFT_TIER_A | DRAFT_TIER_B → QUEUE_REVIEW. Enterprise tier always ESCALATE_HUMAN.
Workflow — do this next
- 01Maintain STAKEHOLDER_MAP with owner.
- 02Drafts-only for 30 days.
- 03approval.flag per recipient before SEND.
- 04Audit log: run_id, approver, recipient hash.
1.4
The strategic intelligence workflow
Competitor monitoring → industry signals → weekly intelligence brief → decision-relevant summary
Key takeaway
STRAT_INTEL_v1: extends Ch 4 multi-source intel — competitor diff + industry RSS allowlist → decision-focused weekly brief.
Why this matters
Leaders need 'so what for our strategy' not news clippings — blueprint forces decision section.
Weekly Sunday 8pm: COMPETITOR_SCAN (parallel per entity) → INDUSTRY_SCAN → JOIN → SYNTH_STRATEGY_v1. Required sections: Moves, Implications, Recommended decisions, Unknowns. Citations mandatory for external claims.
Workflow — do this next
- 01Cap competitors at 8; allowlist sources.
- 02Diff-based scans save tokens.
- 03Route to exec review queue Monday AM.
- 04Quarterly prune low-signal sources.
Ready-to-use artifacts
Complete templates — paste directly into your AI tool or automation workflow.
STRAT_INTEL weekly brief template
## Competitor moves (cited) ## Industry signals ## Implications for us ## Decisions to consider ## Unknowns / gaps
1.5
The performance review workflow
Data inputs → narrative generation → draft review document → manager review queue
Key takeaway
PERF_REVIEW_DRAFT_v1: structured inputs only — goals JSON, feedback bullets, metrics — never invent ratings; managers own final prose.
Why this matters
HR-sensitive workflow demands restricted data class and T3 review — blueprint encodes non-goals.
Inputs: employee_goals.json, 360_bullets.md, team_metrics.csv (anonymised aggregates). GENERATE_REVIEW_DRAFT fills template sections — does not assign performance ratings. Output to confidential staging; access ACL per manager.
Workflow — do this next
- 01Legal/HR sign-off on template + data classes.
- 02No auto-publish to HRIS.
- 03Manager edits all narrative before file.
- 04Delete staging per retention policy.
1.6
The decision memo workflow
Decision context → options analysis → recommendation → formatted memo → distributed for input
Key takeaway
DECISION_MEMO_v1: context pack in → OPTIONS_ANALYSIS → RECOMMENDATION with explicit assumptions → memo.md — circulate for input, not decision by automation.
Why this matters
Decision memos fail when options are vague — blueprint requires 2–4 options with tradeoff table.
Trigger: manual or decision_context/ drop. Skills: INGEST_CONTEXT → OPTIONS_v1 (min 2 options, pros/cons/risks) → RECOMMEND_v1 (states assumptions) → FORMAT_MEMO_v1. Distribute to REVIEW_QUEUE; collecting input is human.
Workflow — do this next
- 01Define DECISION_MEMO_TEMPLATE.
- 02Require assumptions section.
- 03No auto-approve recommendation.
- 04Archive memos with run_id for audit.
Ready-to-use artifacts
Complete templates — paste directly into your AI tool or automation workflow.
Decision memo template
## Context
## Options (table: option | pros | cons | risk)
## Recommendation + assumptions
## Open questions
## Input requested by {date}1.7
The travel and meeting preparation workflow
Calendar → pre-read assembly → briefing document → delivered before each meeting
Key takeaway
MEETING_PREP_v1: poll calendar 45–75 min ahead → pull Drive docs by attendee/project heuristics → PREP_DOC per meeting — DM or folder delivery.
Why this matters
Extends Ch 4 §2.3 with full Skill spec and validation for executive calendar density.
Every 30m: CAL_UPCOMING → DRIVE_MATCH_v1 → PREP_ASSEMBLE_v1. Skip all-day/focus. Include: goal, attendees, linked files, last decisions, suggested questions. VALIDATE: ≥1 linked file or explicit 'no files found'.
Workflow — do this next
- 01Tune Drive search heuristics per exec.
- 02Pilot DM-to-self 2 weeks.
- 03Cap 6 preps per scan — overflow digest.
- 04Exclude confidential external titles from auto-search.
1.8
The week-in-review workflow
Activity log + outputs + metrics → weekly summary → distributed to team and stakeholders
Key takeaway
WIR_v1: aggregate run manifests + key metrics + human bullet input → WEEKLY_LEADER_SUMMARY — Friday 4pm to staging, Monday optional auto-digest after approval.
Why this matters
Closes the loop — leaders see what Cowork did and what the team shipped in one narrative.
Inputs: COWORK_RUN_MANIFESTS week slice, KPI_CSV, leader_bullets.md. AGGREGATE_WIR → NARRATE_WIR_v1. Sections: Wins, Metrics, Automation summary, Next week focus. T2 review before team Slack post.
Workflow — do this next
- 01Auto-collect manifests from run history export.
- 02Leader adds 3 bullets before Friday cron.
- 03Review queue Friday PM; post Monday if approved.
Ready-to-use artifacts
Complete templates — paste directly into your AI tool or automation workflow.
WEEK_IN_REVIEW blueprint
AGGREGATE_WIR → NARRATE_WIR → VALIDATE → REVIEW → (approve) SLACK_POST Schedule: Fri 16:00 | HITL: T2
Concept 2
Operations & Process Workflows
The workflows that run business operations — document processing, reporting, and process automation blueprints
2.1
The document processing pipeline
Folder monitoring → extraction → classification → routing → filing → notification
Key takeaway
DOC_PIPELINE_v1: canonical inbox blueprint — event trigger, classify, extract, route, notify — idempotent move-to-processed.
Why this matters
Highest-ROI ops pattern; this is the reference implementation of Ch 3 §2.2 with full Skill names.
Watch ~/Ops/inbox/ → CLASSIFY_DOC_v1 → branch by type → EXTRACT_* → ROUTE_v1 (Finance|Legal|Ops paths) → NOTIFY_v1. manifest per file; failed stay with ERROR sidecar.
Workflow — do this next
- 01Deploy inbox + processed + failed folders.
- 02One extract Skill per MIME family.
- 03Debounce 90s; max 5 files/run.
- 04Weekly failed-folder triage.
Ready-to-use artifacts
Complete templates — paste directly into your AI tool or automation workflow.
DOC_PIPELINE blueprint
WATCH → CLASSIFY → [EXTRACT_PDF|CSV|DOCX] → ROUTE → NOTIFY
Idempotency: move to processed/{run_id}/
QA: classification confidence < 0.7 → review2.2
The contract review workflow
Contract upload → clause extraction → risk flagging → summary → review queue
Key takeaway
CONTRACT_REVIEW_v1: extract parties, term, renewal, liability, indemnity → RISK_FLAG against playbook → summary for legal — never legal advice, always review queue.
Why this matters
Legal workflows need explicit non-goals and restricted classification — blueprint documents both.
Trigger: PDF in ~/Legal/inbox/. EXTRACT_CONTRACT_v1 → CLAUSE_MAP_v1 → RISK_FLAG_v1 (known risky patterns) → LEGAL_SUMMARY_v1. Output: contract.json + summary.md + flags[]. Route ALL to legal review queue — T3.
Workflow — do this next
- 01Legal owns RISK_PATTERN library.
- 02Disclaimer in every summary output.
- 03No auto-approve or auto-sign.
- 04Retention per legal policy.
2.3
The vendor management workflow
Vendor inputs → evaluation synthesis → comparison matrix → recommendation → filed output
Key takeaway
VENDOR_EVAL_v1: normalise vendor proposals to schema → COMPARE_MATRIX → RECOMMENDATION with criteria weights from Context — filed to ~/Procurement/evals/.
Why this matters
Procurement needs comparable columns — blueprint enforces VENDOR_SCHEMA before synthesis.
Inputs: proposals in ~/Procurement/inbox/{rfp_id}/. NORMALISE_VENDOR_v1 per file → JOIN → MATRIX_v1 → NARRATIVE_REC_v1. Criteria weights in evaluation_rubric.json — human-set, not model-invented.
Workflow — do this next
- 01Define VENDOR_SCHEMA + rubric.
- 02Golden test with 2 past RFPs.
- 03Review queue before procurement lead sign-off.
- 04Archive matrix + manifest per RFP.
Ready-to-use artifacts
Complete templates — paste directly into your AI tool or automation workflow.
VENDOR_SCHEMA fields
vendor_name, pricing_model, term_months, sla, security_certs, implementation_weeks
2.4
The compliance monitoring workflow
Regulatory inputs → change detection → impact assessment → action items → assigned
Key takeaway
COMPLIANCE_MONITOR_v1: diff regulatory feeds → IMPACT_ASSESS per policy map → ACTION_ITEMS with owner hints — assign is human.
Why this matters
Compliance automation assists triage — does not certify compliance.
Daily: REG_FETCH (allowlist) → DIFF vs prior → MATCH_POLICY_MAP_v1 → IMPACT_v1 → ACTIONS.md. Each action: summary, suggested owner role, deadline suggestion, source citation. Route to compliance review queue.
Workflow — do this next
- 01Maintain POLICY_MAP.json with compliance owner.
- 02Citations required on every change item.
- 03Monthly false-positive review.
- 04No auto-close of compliance tickets.
2.5
The project status workflow
Input aggregation → status synthesis → formatted report → stakeholder distribution → archived
Key takeaway
PROJECT_STATUS_v1: collect status md/csv from fixed project folders → SYNTH_STATUS → REPORT_PROJECT_v1 — weekly cron, archive by project/week.
Why this matters
PMO reporting is repetitive and structured — ideal Cowork blueprint.
Monday 7am: scan ~/Projects/*/status_input/ → AGGREGATE → SYNTH_STATUS_v1 (RAG, blockers, milestones) → FORMAT_REPORT → staging → notify PM leads. VALIDATE: every active project has entry or explicit 'no input'.
Workflow — do this next
- 01Define active project list in Context.
- 02PM submits inputs by Sunday 6pm.
- 03Validation flags missing projects.
- 04Archive to ~/Projects/reports/{iso_week}/.
2.6
The expense and financial review workflow
Input documents → extraction → categorisation → anomaly flagging → review queue
Key takeaway
EXPENSE_REVIEW_v1: receipt PDFs + export CSV → extract → categorise per chart → ANOMALY_FLAG (duplicate, threshold) — finance review before ERP.
Why this matters
Money-moving workflows stop at queue — blueprint repeats Ch 5 gates.
Event: files in ~/Finance/expense_inbox/. EXTRACT_RECEIPT → NORMALISE_EXPENSE → MATCH_POLICY_v1 → ANOMALY_v1. Output: expenses.json + review.md for flagged only. No ERP write in v1.
Workflow — do this next
- 01Chart of accounts in Context.
- 02Anomaly thresholds finance-approved.
- 03T3 for any connector write to accounting system.
- 04Reconcile totals to source CSV in QA.
2.7
The onboarding document workflow
New joiner inputs → personalised onboarding pack → delivered to manager → filed
Key takeaway
ONBOARD_PACK_v1: HR form json + role template → ASSEMBLE_ONBOARD_v1 — personalised schedule, links, first-week checklist — to manager Drive folder.
Why this matters
HR ops is document assembly with personalisation — low risk, high time savings.
Trigger: new_hire.json dropped in ~/HR/onboarding/trigger/. MERGE_TEMPLATE (role-based) → PERSONALISE_v1 → PACK.md + links.json → DRIVE_WRITE manager folder + NOTIFY manager. PII: confidential tier.
Workflow — do this next
- 01Maintain role templates per department.
- 02HR triggers via form export json.
- 03Manager review optional T1 notify only.
- 04Delete trigger file after success.
2.8
The KPI reporting workflow
Data inputs → metric calculation → narrative generation → formatted report → distributed on schedule
Key takeaway
KPI_REPORT_v1: INGEST → CLEAN → COMPUTE_METRICS → INTERPRET_FLAGS → REPORT — full Ch 3 §3.8 data-to-decision blueprint with ops naming.
Why this matters
KPI workflow is the analytical pipeline reference most ops teams deploy second after doc inbox.
Condition: kpi_export_ready.flag. Pipeline: INGEST_KPI → CLEAN → ANALYSE → QA reconcile → NARRATE → DISTRIBUTE to ~/Reports/kpi/{date}/ + Slack digest. SLI: QA PASS by 7am Monday.
Workflow — do this next
- 01METRIC_DEFINITIONS signed by leadership.
- 02Golden test vs spreadsheet.
- 03Separate COMPUTE and NARRATE Skills.
- 04Monthly metric definition audit.
Ready-to-use artifacts
Complete templates — paste directly into your AI tool or automation workflow.
KPI_REPORT pipeline
INGEST → CLEAN → ANALYSE → QA → NARRATE → DISTRIBUTE Artifacts: metrics.json, report.md, QA.json Trigger: condition export_ready.flag
Concept 3
Content & Knowledge Workflows
The workflows that run content operations and knowledge management — production-grade designs for content teams and knowledge workers
3.1
The content production pipeline
Brief → research → outline → draft → edit → format → schedule → publish confirmation
Key takeaway
CONTENT_PIPELINE_v1: stage-gated production — each stage is a Skill; publish requires approval.flag — never auto-publish v1.
Why this matters
Content teams need pipeline clarity — blueprint maps creative process to auditable Skills.
Stages: BRIEF_INGEST → RESEARCH_v1 → OUTLINE_v1 → DRAFT_v1 → EDIT_v1 (style card) → FORMAT_v1 → SCHEDULE_STAGING → (T3) PUBLISH_CONFIRM. Handoff manifest between stages; regression fixtures per content type.
Workflow — do this next
- 01Define style card per channel.
- 02Human gate after DRAFT and before PUBLISH.
- 03Store all stages under ~/Content/{project_id}/.
- 04ROI: track mins saved vs review mins.
Ready-to-use artifacts
Complete templates — paste directly into your AI tool or automation workflow.
CONTENT_PIPELINE stages
brief → research → outline → draft → edit → format → schedule → (approve) publish HITL: T2 after draft, T3 before publish
3.2
The research synthesis workflow
Source inputs → extraction → comparison → synthesis → cited output → filed
Key takeaway
RESEARCH_SYNTH_v1: multi-source ingest → EXTRACT per source → COMPARE_MATRIX → SYNTH with numbered citations — filed to ~/Research/{topic}/.
Why this matters
Research without citations is unusable in production — blueprint mandates Sources section.
Inputs: URLs list, PDF drops, internal notes. FETCH → EXTRACT → COMPARE → SYNTH_RESEARCH_v1. Output: synthesis.md + sources.bib.json. Web fetches allowlisted; store retrieval dates.
Workflow — do this next
- 01Allowlist domains in Skill Context.
- 02Require min 3 sources for publish to staging.
- 03QA: every claim maps to source id.
- 04Archive raw fetches 7d per policy.
3.3
The knowledge base maintenance workflow
New inputs → extraction → classification → integration into knowledge base → change log
Key takeaway
KB_MAINTAIN_v1: classify topic → extract facts → propose KB patch (diff format) → human merge — Cowork proposes, human approves KB write.
Why this matters
Fully automated KB writes drift; propose-and-merge is the safe pattern.
Watch ~/KB/inbox/ → CLASSIFY_KB_TOPIC → EXTRACT_FACTS_v1 → PROPOSE_PATCH_v1 (markdown diff) → REVIEW_QUEUE. On approve: optional Notion/MCP write or manual paste. CHANGELOG.md append per run_id.
Workflow — do this next
- 01Define KB taxonomy.
- 02Patch format: diff only, not full replace.
- 03Owner approves merges weekly.
- 04Version KB exports monthly for regression.
3.6
The competitive intelligence workflow
Competitor monitoring → change detection → synthesis → weekly brief → distributed
Key takeaway
COMP_INTEL_v1: same core as §1.4 with content-team delivery — formatted for internal wiki + Slack #competitive.
Why this matters
Content and strategy teams share intel machinery — blueprint notes audience-specific formatting Skill.
COMP_SCAN parallel → DIFF → SYNTH_COMP_v1 → FORMAT_WIKI_v1 + SLACK_DIGEST_v1. Emphasise product/marketing implications section. Schedule Monday 7am.
Workflow — do this next
- 01Share scan Skills with STRAT_INTEL where possible.
- 02Audience format Skills separate from SYNTH.
- 03Wiki write to staging; human promote.
3.7
The learning digest workflow
Reading list inputs → summary extraction → key insight synthesis → formatted digest → distributed
Key takeaway
LEARNING_DIGEST_v1: reading list csv/urls → EXTRACT per item → SYNTH_INSIGHTS → digest.md — team L&D distribution on schedule.
Why this matters
Teams curate more than they read — digest makes reading lists actionable.
Biweekly: READING_LIST ingest → BATCH_EXTRACT (max 10 items) → SYNTH_DIGEST_v1. Sections: One-line thesis per item, Cross-cutting themes, Suggested actions. Cite each item.
Workflow — do this next
- 01Curator maintains reading_list.csv.
- 02Cap items per run for token budget.
- 03Distribute to Slack #learning or email draft.
- 04Archive digests by period.
3.8
The meeting notes-to-knowledge workflow
Meeting transcript → action extraction → decision logging → knowledge base update → notifications sent
Key takeaway
NOTES_TO_KB_v1: transcript → ACTIONS + DECISIONS extract → PROPOSE_KB_ENTRIES → notify owners — actions do not auto-create tickets without approval.
Why this matters
Meeting notes are the richest knowledge feed — blueprint structures extract → propose → notify.
Drop transcript in ~/Meetings/inbox/. EXTRACT_ACTIONS_v1 → EXTRACT_DECISIONS_v1 → PROPOSE_KB_v1 → NOTIFY_OWNERS_v1 (Slack DM template with action list). Ticket creation optional v2 with approval.
Workflow — do this next
- 01Transcript format normalised (vtt/md).
- 02Owner assignment suggestions — human confirms.
- 03KB proposals to review queue.
- 04Notify action owners — no auto-assign in v1.
Ready-to-use artifacts
Complete templates — paste directly into your AI tool or automation workflow.
NOTES_TO_KB pipeline
TRANSCRIPT → EXTRACT_ACTIONS | EXTRACT_DECISIONS
→ PROPOSE_KB → REVIEW → NOTIFY
Output: actions.json, decisions.json, kb_patch.diffReady-to-use artifacts
Complete templates — paste directly into your AI tool or automation workflow.
Blueprint index (24 workflows)
Quick reference — section ID to blueprint name.
Exec: EXEC_BRIEF, BOARD_PACK, STAKEHOLDER_COMMS, STRAT_INTEL, PERF_REVIEW, DECISION_MEMO, MEETING_PREP, WIR Ops: DOC_PIPELINE, CONTRACT_REVIEW, VENDOR_EVAL, COMPLIANCE_MONITOR, PROJECT_STATUS, EXPENSE_REVIEW, ONBOARD_PACK, KPI_REPORT Content: CONTENT_PIPELINE, RESEARCH_SYNTH, KB_MAINTAIN, NEWSLETTER, SOCIAL_MULTI, COMP_INTEL, LEARNING_DIGEST, NOTES_TO_KB
Blueprint adaptation checklist
Steps to clone any Ch 6 blueprint to your environment.
[ ] Copy pipeline spec [ ] Map paths + connector allowlists [ ] PATH_MAPPING sandbox → prod [ ] 5-run test + regression [ ] Production readiness gate [ ] Schedule + monitor 3 cycles
EXEC_MORNING_BRIEF blueprint
Skills: CAL_FETCH | GMAIL_FETCH | SLACK_READ | NEWS_SCAN
→ JOIN → PRIORITISE_v1 → BRIEF_EXEC_v1 → VALIDATE → EMIT
Schedule: 06:30 | HITL: T2 | SLA: brief by 07:45
Output: ~/Briefs/exec/{date}/brief.mdBOARD_PACK pipeline spec
INGEST → EXTRACT_* → SYNTH_BOARD → QA → FORMAT → REVIEW_QUEUE
→ (T3) DRIVE_DISTRIBUTE
Freeze: no Skill version bumps 7d before boardSTRAT_INTEL weekly brief template
## Competitor moves (cited) ## Industry signals ## Implications for us ## Decisions to consider ## Unknowns / gaps
Decision memo template
## Context
## Options (table: option | pros | cons | risk)
## Recommendation + assumptions
## Open questions
## Input requested by {date}WEEK_IN_REVIEW blueprint
AGGREGATE_WIR → NARRATE_WIR → VALIDATE → REVIEW → (approve) SLACK_POST Schedule: Fri 16:00 | HITL: T2
DOC_PIPELINE blueprint
WATCH → CLASSIFY → [EXTRACT_PDF|CSV|DOCX] → ROUTE → NOTIFY
Idempotency: move to processed/{run_id}/
QA: classification confidence < 0.7 → reviewVENDOR_SCHEMA fields
vendor_name, pricing_model, term_months, sla, security_certs, implementation_weeks
KPI_REPORT pipeline
INGEST → CLEAN → ANALYSE → QA → NARRATE → DISTRIBUTE Artifacts: metrics.json, report.md, QA.json Trigger: condition export_ready.flag
CONTENT_PIPELINE stages
brief → research → outline → draft → edit → format → schedule → (approve) publish HITL: T2 after draft, T3 before publish
NOTES_TO_KB pipeline
TRANSCRIPT → EXTRACT_ACTIONS | EXTRACT_DECISIONS
→ PROPOSE_KB → REVIEW → NOTIFY
Output: actions.json, decisions.json, kb_patch.diffCoE — blueprint library cut time-to-production 70%
A 45-person company's Cowork CoE spent 3 weeks per new workflow inventing pipelines from scratch — inconsistent gates, duplicate Skills, frustrated domain owners.
Before
11 custom one-offs, 4 failed schedules, no shared naming, executives still asking 'can Cowork do X?'
After
Chapter 6 blueprint library in ops wiki; adaptation checklist mandatory; exec brief + doc pipeline + KPI report as standard trio; new workflows clone and ship in 4–5 days.
- Time-to-production → 21 days median to 6 days
- Duplicate Skills → consolidated from 11 overlapping to 6 canonical
- Production incidents → 2 in Q1 to 0 after blueprint QA patterns
- Adoption → 18 live workflows in 2 quarters vs 5 in prior year
What goes wrong
Copying blueprint Skill names without TAR specs — behaviour drift.
Clone full TAR + fixtures per Ch 2 §3.8 share bundle.
Skipping validation because 'blueprint looked production-ready.'
Ch 5 production gate on every adaptation — blueprints assume gates, do not replace them.
Enabling send/publish nodes on day one.
Blueprints default drafts/review — T3 only after pilot period.
24 blueprints attempted at once — CoE overwhelmed.
§3.2 priority matrix — deploy standard trio first, one new blueprint per sprint.
Vetted by Krishna KumarCurator, FactorBeam
Discussion
Discussion coming soon
Shared comments for this playbook are not live yet. When they are, you'll be able to ask questions, share what worked, and see replies from other readers.
3.5
The social content workflow
Source material → multi-format content generation → review queue → scheduled for publishing
Key takeaway
SOCIAL_MULTI_v1: one source brief → PARALLEL draft LinkedIn, X, thread variants → JOIN → review queue — schedule posts only with per-post approval.flag.
Why this matters
Multi-format from one source saves time; per-channel style cards prevent voice bleed.
Input: source.md or blog URL fetch. BRANCH: DRAFT_LINKEDIN | DRAFT_X | DRAFT_THREAD → JOIN_REVIEW. Style card per platform. Publishing connector optional v2 — v1 exports to ~/Social/scheduled/.
Workflow — do this next