workflow-orchestrator
Chains positioning, competitive, GTM strategy, campaign briefs, stakeholder maps and retros into one coherent PMM program with a master document and brain updates. Trigger on: "run full GTM workflow", "run a full launch", "launch [product] end to end", "positioning refresh", "qua
Install
npx skills add https://github.com/stefanoskarakasis/Product-Marketing-Skills/tree/main/pmm-go-to-market/skills/workflow-orchestrator
claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install stefanoskarakasis-product-marketing-skills@llmmart
git clone https://github.com/stefanoskarakasis/Product-Marketing-Skills.git
The skills CLI installs just this skill, for any of its supported agents. Claude Code installs the whole stefanoskarakasis/product-marketing-skills collection as a plugin from our marketplace. Git is the plain clone.
Skill manifest
PMM Workflow Orchestrator
Chains your PMM skills into complete end-to-end programs. Manages state across skills, verifies coherence between outputs, writes back to brain, and produces one master program document the team can act from.
Not a shortcut for running one skill. A program manager for multi-skill PMM work where consistency across outputs — positioning, competitive, GTM, stakeholder — matters as much as each individual output.
Trigger
- When: Any request for a multi-step PMM program spanning two or more skills, or any named workflow type below.
- Not for: Single-skill tasks — route directly to the relevant skill instead:
go-to-market-strategyfor tier + brief only,gtm-motionsfor motion selection only,positioning-messagingfor positioning work only,pre-mortemfor risk analysis only,retrofor post-launch review only. - Example prompts:
- "Run a full launch workflow for our analytics dashboard"
- "We're doing a positioning refresh — orchestrate the full program"
- "Run our quarterly PMM cycle"
- "New market entry into healthcare — full program"
- "I just joined as PMM — audit the current state"
- "Run a competitive intelligence program on our top 3 competitors"
Inputs
- Args: Workflow type (named or described) + initiative context. See Commands for supported workflow types. Free-form description also accepted — orchestrator will infer the closest workflow and confirm before running.
- Defaults: If workflow type is ambiguous, ask before starting. Never infer and execute silently on a multi-skill program.
- Context keys:
/foundation/brain.md— required. All sections.- Brain contract:
- Reads: Sections 1–6 — checks completeness and staleness before routing.
- Writes: Section 3 (after positioning or alternatives refresh, via
alternatives-maporpositioning-messaging), Section 6 (after new proof points confirmed, viaproof-points). - Never writes to: Section 1, Section 2, Section 4, Section 5 (those require
dedicated skills:
product-marketing-context). The orchestrator itself never writes brain content directly in any section — every write above is performed by the named downstream skill and only confirmed here.
- Staleness thresholds:
- Section 3 (Alternatives & Positioning) > 6 months → flag for refresh before any launch workflow.
- Section 6 (Proof Points) containing unresolved
[NEEDS PROOF]/[NEEDS APPROVAL]entries → flag for aproof-pointsAudit pass before any workflow that cites claims externally (sales enablement, launch brief).
Pre-flight
- Load
/foundation/brain.md. Read all sections silently. - If brain missing: stop before the charter and surface:
"Brain file not found at
/foundation/brain.md. Runbeachhead-segmentfirst (same plugin, it builds a quick brain in about 3 minutes), then come back. The orchestrator needs your ICP and positioning to route skills coherently.product-marketing-context(a separate plugin) builds the full brain." Unlike individual skills, the orchestrator does not run on nothing. A program of disconnected outputs is worse than no program. - Audit brain sections before confirming workflow:
- Section 3 > 6 months old → "Positioning is [N] months old. This workflow will include a positioning refresh step."
- Section 6 has unresolved
[NEEDS PROOF]/[NEEDS APPROVAL]entries → "Proof points registry has [N] unconfirmed claims. Will runproof-pointsAudit mode before any step that cites claims externally." - Any section 🔴 Placeholder → flag: "Section is Placeholder. This limits [specific skill] output quality. Recommend completing before running."
- Confirm Program Charter with user before invoking any skill. No skill runs without explicit go-ahead.
Steps
Step 1: Intake and Program Definition
Ask in one message. Adapt based on which workflow type is named.
"Before I start the program, I need to lock a few things:
What's the initiative? (One sentence — what are we launching, refreshing, or auditing?)
Who is the DRI? (The one person who owns this program's outcomes)
What's the primary success metric and timeline? (The number we're building toward, and when)
Which skills have already run? (Don't re-run what's current — I'll check brain staleness and confirm what needs to run vs. what can be skipped)"
Reflect back as a Program Charter before proceeding:
PROGRAM CHARTER
───────────────────────────────────────
Initiative: [name]
Workflow type: [named workflow]
DRI: [name]
Primary metric: [metric + target]
Timeline: [start → end]
Brain status: [Section staleness summary]
Skills to run: [ordered list]
Skills to skip: [list with reason]
───────────────────────────────────────
Confirm to start? [Y/N]
Do not proceed until confirmed.
Step 2: Brain Staleness Check
For each brain section relevant to the workflow, assess:
| Section | Last updated | Status | Action |
|---|---|---|---|
| 3 — Positioning | [date or unknown] | 🟢 Current / 🟡 Aging / 🔴 Stale | Run / Skip / Flag |
| 4 — Competitive | [date or unknown] | 🟢 / 🟡 / 🔴 | Run / Skip / Flag |
Surface as part of the Program Charter — do not run silently.
Step 3: Skill Sequencing
Determine the ordered skill chain for the confirmed workflow. Surface the sequence before running:
"Here's the sequence for this program:
- [Skill A] — [what it produces and why it's first]
- [Skill B] — [what it reads from A and what it adds] ... Each skill output feeds the next. I'll flag if a downstream skill can't proceed due to a gap in an upstream output."
Step 4: Skill Execution
Invoke each skill in sequence. Between each skill:
- Confirm output with user before passing state to next skill.
- Extract relevant outputs and update the Program State document.
- Flag any coherence issue immediately before running the next skill.
Coherence checks between skills:
- Positioning → GTM strategy: do channel recommendations match the positioned segment?
- GTM motions → GTM strategy: does the Channel Strategy table reflect the selected motion stack, not a leftover tier-default?
- GTM strategy → Stakeholder maps: does the tier align with stakeholder weight?
- Competitive → GTM strategy: does the attack angle match the battlecard?
- Personas → Proof points: does the proof point set address the persona's primary objection?
If coherence fails: surface the conflict, ask which output to trust, re-run the downstream skill with corrected input.
Learning Close per chained skill: Every T1/T2 skill invoked in this
sequence runs its own Learning Close step as the last step of its own
## Steps — logging one row to /context/skill-sessions.md before control
returns to the orchestrator. Do not skip or short-circuit a skill's own
Learning Close when chaining; the orchestrator does not log on a skill's
behalf, it only confirms the skill completed all its own steps, including
that one.
Step 5: Brain Update
After each skill produces a confirmed output:
- Positioning or alternatives refresh: Update Section 3 with new statement and timestamp (via
positioning-messagingoralternatives-map). - New proof points: Update Section 6 with approved new claims (via
proof-points— Add mode for a single sourced claim, Extract mode if pulling from a deck or battlecard, Audit mode to re-flag stale or unsupported entries).
Surface each write:
"Updating brain Section [X] with [what]. Confirm? [Y/N]"
If the user wants launch or retro history tracked somewhere, ask where they'd like it kept — this skill doesn't maintain its own log automatically.
Step 6: Master Program Document
Compile all skill outputs into one master document:
# [Program Name] — Master Document
**Workflow type:** [named]
**DRI:** [name]
**Period:** [start → end]
**Primary metric:** [metric + target]
**Brain last updated:** [date]
## Program Charter
[From Step 1]
## Skill Outputs
### [Skill A] — [date run]
[Confirmed output or summary]
### [Skill B] — [date run]
[Confirmed output or summary]
## Coherence Check Results
[Conflicts surfaced and how resolved]
## Brain Updates Made
[Sections updated with timestamps]
## Open Items
[Unresolved gaps — owner and due date]
## Next Program Trigger
[When to run the next cycle]
If the user wants this document saved somewhere, ask where.
Step 7: Close
- Run
/coherenceacross all completed outputs — resolve any remaining conflicts. - Surface next program trigger date:
- Full launch → retro in 90 days
- Quarterly cycle → next cycle in 13 weeks
- Positioning refresh → re-audit in 6 months
- Competitive program → re-run in 3 months
- Append one entry to
/context/skill-sessions.mdfor the orchestration itself — separate from, and in addition to, each chained skill's own Learning Close row (per Step 4's delegation, unchanged):
type: execution
skill: workflow-orchestrator
session_date: [YYYY-MM-DD]
pattern: [one falsifiable statement about the orchestration — workflow type,
number of skills chained, whether coherence check passed clean or needed
conflict resolution, or "none"]
source: [surprised / wrong / missing / N/A]
Write this row directly — do not ask the user for permission.
Supported Workflows
1. Full Product Launch
Trigger: "run full launch workflow", "launch [product] end to end"
go-to-market-strategy— tier + strategy briefgtm-motions— select acquisition motion stack, fold into Channel Strategypositioning-messaging(if Section 3 stale or new angles needed)gaccs-brief— campaign briefstakeholder-maps— internal alignment mappre-mortem— risk analysis before committing- Set T+90 retro trigger →
retro
2. Positioning Refresh
Trigger: "positioning refresh", "our messaging is stale", "update positioning"
positioning-messaging(AUDIT mode → full BUILD if audit score < 70)
3. Competitive Intelligence Program
Trigger: "competitive program", "build battlecards", "competitive deep-dive"
alternatives-map(frompmm-positioning; skip if brain Section 3 already names the competitors with why-chosen and why-left)competitive-profile(from thepmm-competitive-intelligenceplugin; skip if current profiles exist) — fixes the roster of up to five and builds their profilescompetitive-battlecard(same plugin; if it is not installed, say so and stop) — one card per competitor the user names, with a checkpoint between competitorsproof-points— register the claims the cards listed as[NEEDS PROOF]
4. Quarterly PMM Cycle
Trigger: "quarterly PMM cycle", "Q[X] refresh", "quarterly review"
retro— debrief all launches from prior quarterpositioning-messaging(AUDIT mode)pmm-okrs— set next quarter OKRs
5. New Market Entry Program
Trigger: "enter new market", "expand to [segment]", "new vertical"
beachhead-segment— score and confirm the target segmentgtm-motions— select acquisition motion stack for the confirmed segmentpositioning-messaging— positioning for new segmentgo-to-market-strategy— tier + strategy for market entrygaccs-brief— campaign briefpre-mortem— risk analysis before committing
6. Competitive Response (Fast)
Trigger: "competitive response to [competitor]", "they just launched [X]"
competitive-battlecard(from thepmm-competitive-intelligenceplugin; if it is not installed, say so and stop) — for the named competitor only, not all competitors. Pass along the deal context the user gave (role, buyer, stage); the skill asks once for the rest.- Offer, never run:
proof-pointsfor any claim the card listed as[NEEDS PROOF]
7. Full PMM Onboarding / Audit
Trigger: "I just joined as PMM", "PMM audit", "what's our current state"
- Audit brain Sections 1–6 — completeness, staleness, gaps
positioning-messaging(AUDIT mode only)- Produce Current State Report with prioritised gaps
Note: Read-only. No brain writes under any circumstance.
8. Weekly Opportunities
Trigger: "weekly opportunities", "what changed this week", "weekly PMM read"
signal-monitor(from thepmm-intelligenceplugin; if it is not installed, say so and stop) — at most five opportunities from the signals the user names- For each opportunity the user wants to act on, route to the one skill it names (
alternatives-map,positioning-messagingorpmm-okrs), with a checkpoint before it runs
Note: The read itself writes only its dated list in /context/signals/. Any brain write is performed by the downstream skill after the user confirms.
Commands
/run [workflow-type]
Start a named workflow. Runs Steps 1–7 with checkpoints.
/run full-launch
/run positioning-refresh
/run competitive-program
/run quarterly-cycle
/run market-entry
/run competitive-response [competitor name]
/run audit
/run weekly-opportunities
/status
Show current program state: skills completed, skills pending, next skill, brain updates made.
/skip [skill-name]
Skip a skill in the current sequence. Requires reason. Orchestrator flags downstream impact and requests confirmation before skipping.
/coherence
Run a coherence check across all completed outputs. Flags positioning ↔ GTM, competitive ↔ GTM, and personas ↔ proof point conflicts with specific fixes.
/compile
Compile the Master Program Document from all confirmed skill outputs. Can be run mid-program or at end.
/next
Show what skill runs next, what it reads from prior outputs, what it produces, and estimated session time.
Outputs
- Files written:
/foundation/brain.md— Sections 3 and 6 updated on confirmation after relevant skills complete (Section 3 viapositioning-messagingoralternatives-map; Section 6 viaproof-points)./context/skill-sessions.md— one row per chained skill's own Learning Close, plus one row for the orchestration itself (Step 7). Master Program Document is produced in chat; ask the user where, if anywhere, to save it. - Chat output format: Program Charter → sequential skill outputs with checkpoints → coherence check results → Master Program Document. Each skill output is clearly delimited with skill name and date.
- External side effects: None beyond confirmed brain writes.
Verification
Runs at program close, before Master Program Document is delivered.
| Check | Standard | Pass = |
|---|---|---|
| Program Charter confirmed | User explicitly confirmed charter before any skill ran | Yes |
| Each skill confirmed | Every output confirmed by user before next skill ran | Yes |
| Coherence check run | /coherence run after all skills; conflicts surfaced and resolved | Yes |
| Brain writes confirmed | Every brain write shown and confirmed before executing | Yes |
| Master document complete | All required sections present with no orphaned outputs | Yes |
| Next trigger surfaced | Next program cycle date recommended at close | Yes |
Do Not Use For
- go-to-market-strategy — when you only need a tier assignment and GTM brief, not a multi-skill program. Use that skill directly.
- positioning-messaging — when you only need positioning or messaging work. Use that skill directly.
- retro — for a single post-launch retrospective outside a larger program cycle. Use that skill directly.
- pre-mortem — for standalone risk analysis before a launch. Use that skill directly, then bring the output into a full launch workflow if needed.
Operating Rules
- A brain is required, a quick one is enough. The orchestrator does not run without
/foundation/brain.md; the Quick-Brain frombeachhead-segmentsatisfies it. Disconnected skill outputs without shared context are worse than no program. - Program Charter before execution. No skill runs before the charter is confirmed. Starting without alignment is the most common reason programs produce wasted output.
- Checkpoint after every skill. User confirms each output before the next skill runs. Silent chaining is not permitted — it produces coherence failures that are expensive to unwind.
- Coherence over speed. If a conflict surfaces between outputs, stop and resolve it before proceeding. A fast program with incoherent outputs is not a program.
- Staleness must be surfaced. If any brain section relevant to the workflow is stale, say so in the charter. The user decides whether to refresh — consciously.
- Brain writes require confirmation. Never write to brain without explicit user go-ahead. Show what will be written, then wait.
- Skip requires downstream impact disclosure. If the user skips a skill, tell them exactly what the next skill loses by not having that output.
- Read-only audits write nothing. The Full PMM Onboarding / Audit workflow does not write to brain under any circumstance.
- One DRI per program. If no DRI is named at intake, ask before confirming the charter. A program with no single owner has no owner.
- Next program trigger at close. Every completed program ends with a recommended date for the next cycle.
Quality Gate
Runs at program close, before Master Program Document is delivered.
| Check | Standard | Pass = |
|---|---|---|
| Program Charter confirmed | User explicitly confirmed charter before skills ran | Yes |
| All skills accounted for | Each skill either ran or was explicitly skipped with reason | Yes |
| Checkpoint for each skill | Each skill output confirmed before next skill ran | Yes |
| Coherence check run | /coherence run after all skills — conflicts surfaced and resolved | Yes |
| Brain writes confirmed | Every brain write shown to user and confirmed | Yes |
| Master document complete | All required sections present in master document | Yes |
| Open items documented | Unresolved gaps have named owner and due date | Yes |
| Next trigger surfaced | Next program cycle date recommended at close | Yes |
| No orphaned outputs | Every skill output referenced in master document | Yes |
| DRI named | Program Charter includes a named DRI | Yes |
Files (product-marketing-skills)
-
evals
-
workflow-orchestrator.eval.md 9.4 KB
Loading…
-
-
SKILL.md 19 KB
--- name: workflow-orchestrator version: 2.6.2 description: > Chains positioning, competitive, GTM strategy, campaign briefs, stakeholder maps and retros into one coherent PMM program with a master document and brain updates. Trigger on: "run full GTM workflow", "run a full launch", "launch [product] end to end", "positioning refresh", "quarterly PMM cycle", "competitive program", "new market entry program", "full PMM onboarding". metadata: author: Stefanos Karakasis context: brain-dependent quality_gate: true last_updated: 2026-11-12 --- # PMM Workflow Orchestrator Chains your PMM skills into complete end-to-end programs. Manages state across skills, verifies coherence between outputs, writes back to brain, and produces one master program document the team can act from. Not a shortcut for running one skill. A program manager for multi-skill PMM work where consistency across outputs — positioning, competitive, GTM, stakeholder — matters as much as each individual output. --- ## Trigger - **When:** Any request for a multi-step PMM program spanning two or more skills, or any named workflow type below. - **Not for:** Single-skill tasks — route directly to the relevant skill instead: `go-to-market-strategy` for tier + brief only, `gtm-motions` for motion selection only, `positioning-messaging` for positioning work only, `pre-mortem` for risk analysis only, `retro` for post-launch review only. - **Example prompts:** - "Run a full launch workflow for our analytics dashboard" - "We're doing a positioning refresh — orchestrate the full program" - "Run our quarterly PMM cycle" - "New market entry into healthcare — full program" - "I just joined as PMM — audit the current state" - "Run a competitive intelligence program on our top 3 competitors" --- ## Inputs - **Args:** Workflow type (named or described) + initiative context. See Commands for supported workflow types. Free-form description also accepted — orchestrator will infer the closest workflow and confirm before running. - **Defaults:** If workflow type is ambiguous, ask before starting. Never infer and execute silently on a multi-skill program. - **Context keys:** - `/foundation/brain.md` — required. All sections. - **Brain contract:** - Reads: Sections 1–6 — checks completeness and staleness before routing. - Writes: Section 3 (after positioning or alternatives refresh, via `alternatives-map` or `positioning-messaging`), Section 6 (after new proof points confirmed, via `proof-points`). - Never writes to: Section 1, Section 2, Section 4, Section 5 (those require dedicated skills: `product-marketing-context`). The orchestrator itself never writes brain content directly in any section — every write above is performed by the named downstream skill and only confirmed here. - **Staleness thresholds:** - Section 3 (Alternatives & Positioning) > 6 months → flag for refresh before any launch workflow. - Section 6 (Proof Points) containing unresolved `[NEEDS PROOF]` / `[NEEDS APPROVAL]` entries → flag for a `proof-points` Audit pass before any workflow that cites claims externally (sales enablement, launch brief). --- ## Pre-flight - Load `/foundation/brain.md`. Read all sections silently. - If brain missing: stop before the charter and surface: > "Brain file not found at `/foundation/brain.md`. Run `beachhead-segment` first (same plugin, it builds a quick brain in about 3 minutes), then come back. The orchestrator needs your ICP and positioning to route skills coherently. `product-marketing-context` (a separate plugin) builds the full brain." > Unlike individual skills, the orchestrator does not run on nothing. A program of disconnected outputs is worse than no program. - Audit brain sections before confirming workflow: - Section 3 > 6 months old → "Positioning is [N] months old. This workflow will include a positioning refresh step." - Section 6 has unresolved `[NEEDS PROOF]` / `[NEEDS APPROVAL]` entries → "Proof points registry has [N] unconfirmed claims. Will run `proof-points` Audit mode before any step that cites claims externally." - Any section 🔴 Placeholder → flag: "Section [X] is Placeholder. This limits [specific skill] output quality. Recommend completing before running." - Confirm Program Charter with user before invoking any skill. No skill runs without explicit go-ahead. --- ## Steps ### Step 1: Intake and Program Definition Ask in one message. Adapt based on which workflow type is named. > "Before I start the program, I need to lock a few things: > > 1. **What's the initiative?** (One sentence — what are we launching, refreshing, > or auditing?) > > 2. **Who is the DRI?** (The one person who owns this program's outcomes) > > 3. **What's the primary success metric and timeline?** > (The number we're building toward, and when) > > 4. **Which skills have already run?** (Don't re-run what's current — I'll check > brain staleness and confirm what needs to run vs. what can be skipped)" Reflect back as a Program Charter before proceeding: ```` PROGRAM CHARTER ─────────────────────────────────────── Initiative: [name] Workflow type: [named workflow] DRI: [name] Primary metric: [metric + target] Timeline: [start → end] Brain status: [Section staleness summary] Skills to run: [ordered list] Skills to skip: [list with reason] ─────────────────────────────────────── Confirm to start? [Y/N] ```` Do not proceed until confirmed. --- ### Step 2: Brain Staleness Check For each brain section relevant to the workflow, assess: | Section | Last updated | Status | Action | |---|---|---|---| | 3 — Positioning | [date or unknown] | 🟢 Current / 🟡 Aging / 🔴 Stale | Run / Skip / Flag | | 4 — Competitive | [date or unknown] | 🟢 / 🟡 / 🔴 | Run / Skip / Flag | Surface as part of the Program Charter — do not run silently. --- ### Step 3: Skill Sequencing Determine the ordered skill chain for the confirmed workflow. Surface the sequence before running: > "Here's the sequence for this program: > 1. [Skill A] — [what it produces and why it's first] > 2. [Skill B] — [what it reads from A and what it adds] > ... > Each skill output feeds the next. I'll flag if a downstream skill can't proceed > due to a gap in an upstream output." --- ### Step 4: Skill Execution Invoke each skill in sequence. Between each skill: 1. Confirm output with user before passing state to next skill. 2. Extract relevant outputs and update the Program State document. 3. Flag any coherence issue immediately before running the next skill. **Coherence checks between skills:** - Positioning → GTM strategy: do channel recommendations match the positioned segment? - GTM motions → GTM strategy: does the Channel Strategy table reflect the selected motion stack, not a leftover tier-default? - GTM strategy → Stakeholder maps: does the tier align with stakeholder weight? - Competitive → GTM strategy: does the attack angle match the battlecard? - Personas → Proof points: does the proof point set address the persona's primary objection? If coherence fails: surface the conflict, ask which output to trust, re-run the downstream skill with corrected input. **Learning Close per chained skill:** Every T1/T2 skill invoked in this sequence runs its own Learning Close step as the last step of its own `## Steps` — logging one row to `/context/skill-sessions.md` before control returns to the orchestrator. Do not skip or short-circuit a skill's own Learning Close when chaining; the orchestrator does not log on a skill's behalf, it only confirms the skill completed all its own steps, including that one. --- ### Step 5: Brain Update After each skill produces a confirmed output: - **Positioning or alternatives refresh:** Update Section 3 with new statement and timestamp (via `positioning-messaging` or `alternatives-map`). - **New proof points:** Update Section 6 with approved new claims (via `proof-points` — Add mode for a single sourced claim, Extract mode if pulling from a deck or battlecard, Audit mode to re-flag stale or unsupported entries). Surface each write: > "Updating brain Section [X] with [what]. Confirm? [Y/N]" If the user wants launch or retro history tracked somewhere, ask where they'd like it kept — this skill doesn't maintain its own log automatically. --- ### Step 6: Master Program Document Compile all skill outputs into one master document: ````markdown # [Program Name] — Master Document **Workflow type:** [named] **DRI:** [name] **Period:** [start → end] **Primary metric:** [metric + target] **Brain last updated:** [date] ## Program Charter [From Step 1] ## Skill Outputs ### [Skill A] — [date run] [Confirmed output or summary] ### [Skill B] — [date run] [Confirmed output or summary] ## Coherence Check Results [Conflicts surfaced and how resolved] ## Brain Updates Made [Sections updated with timestamps] ## Open Items [Unresolved gaps — owner and due date] ## Next Program Trigger [When to run the next cycle] ```` If the user wants this document saved somewhere, ask where. --- ### Step 7: Close 1. Run `/coherence` across all completed outputs — resolve any remaining conflicts. 2. Surface next program trigger date: - Full launch → retro in 90 days - Quarterly cycle → next cycle in 13 weeks - Positioning refresh → re-audit in 6 months - Competitive program → re-run in 3 months 3. Append one entry to `/context/skill-sessions.md` for the orchestration itself — separate from, and in addition to, each chained skill's own Learning Close row (per Step 4's delegation, unchanged): ````yaml type: execution skill: workflow-orchestrator session_date: [YYYY-MM-DD] pattern: [one falsifiable statement about the orchestration — workflow type, number of skills chained, whether coherence check passed clean or needed conflict resolution, or "none"] source: [surprised / wrong / missing / N/A] ```` Write this row directly — do not ask the user for permission. --- ## Supported Workflows ### 1. Full Product Launch **Trigger:** "run full launch workflow", "launch [product] end to end" 1. `go-to-market-strategy` — tier + strategy brief 2. `gtm-motions` — select acquisition motion stack, fold into Channel Strategy 3. `positioning-messaging` (if Section 3 stale or new angles needed) 4. `gaccs-brief` — campaign brief 5. `stakeholder-maps` — internal alignment map 6. `pre-mortem` — risk analysis before committing 7. Set T+90 retro trigger → `retro` ### 2. Positioning Refresh **Trigger:** "positioning refresh", "our messaging is stale", "update positioning" 1. `positioning-messaging` (AUDIT mode → full BUILD if audit score < 70) ### 3. Competitive Intelligence Program **Trigger:** "competitive program", "build battlecards", "competitive deep-dive" 1. `alternatives-map` (from `pmm-positioning`; skip if brain Section 3 already names the competitors with why-chosen and why-left) 2. `competitive-profile` (from the `pmm-competitive-intelligence` plugin; skip if current profiles exist) — fixes the roster of up to five and builds their profiles 3. `competitive-battlecard` (same plugin; if it is not installed, say so and stop) — one card per competitor the user names, with a checkpoint between competitors 4. `proof-points` — register the claims the cards listed as `[NEEDS PROOF]` ### 4. Quarterly PMM Cycle **Trigger:** "quarterly PMM cycle", "Q[X] refresh", "quarterly review" 1. `retro` — debrief all launches from prior quarter 2. `positioning-messaging` (AUDIT mode) 3. `pmm-okrs` — set next quarter OKRs ### 5. New Market Entry Program **Trigger:** "enter new market", "expand to [segment]", "new vertical" 1. `beachhead-segment` — score and confirm the target segment 2. `gtm-motions` — select acquisition motion stack for the confirmed segment 3. `positioning-messaging` — positioning for new segment 4. `go-to-market-strategy` — tier + strategy for market entry 5. `gaccs-brief` — campaign brief 6. `pre-mortem` — risk analysis before committing ### 6. Competitive Response (Fast) **Trigger:** "competitive response to [competitor]", "they just launched [X]" 1. `competitive-battlecard` (from the `pmm-competitive-intelligence` plugin; if it is not installed, say so and stop) — for the named competitor only, not all competitors. Pass along the deal context the user gave (role, buyer, stage); the skill asks once for the rest. 2. Offer, never run: `proof-points` for any claim the card listed as `[NEEDS PROOF]` ### 7. Full PMM Onboarding / Audit **Trigger:** "I just joined as PMM", "PMM audit", "what's our current state" 1. Audit brain Sections 1–6 — completeness, staleness, gaps 2. `positioning-messaging` (AUDIT mode only) 3. Produce Current State Report with prioritised gaps **Note:** Read-only. No brain writes under any circumstance. ### 8. Weekly Opportunities **Trigger:** "weekly opportunities", "what changed this week", "weekly PMM read" 1. `signal-monitor` (from the `pmm-intelligence` plugin; if it is not installed, say so and stop) — at most five opportunities from the signals the user names 2. For each opportunity the user wants to act on, route to the one skill it names (`alternatives-map`, `positioning-messaging` or `pmm-okrs`), with a checkpoint before it runs **Note:** The read itself writes only its dated list in `/context/signals/`. Any brain write is performed by the downstream skill after the user confirms. --- ## Commands ### /run [workflow-type] Start a named workflow. Runs Steps 1–7 with checkpoints. ```` /run full-launch /run positioning-refresh /run competitive-program /run quarterly-cycle /run market-entry /run competitive-response [competitor name] /run audit /run weekly-opportunities ```` ### /status Show current program state: skills completed, skills pending, next skill, brain updates made. ### /skip [skill-name] Skip a skill in the current sequence. Requires reason. Orchestrator flags downstream impact and requests confirmation before skipping. ### /coherence Run a coherence check across all completed outputs. Flags positioning ↔ GTM, competitive ↔ GTM, and personas ↔ proof point conflicts with specific fixes. ### /compile Compile the Master Program Document from all confirmed skill outputs. Can be run mid-program or at end. ### /next Show what skill runs next, what it reads from prior outputs, what it produces, and estimated session time. --- ## Outputs - **Files written:** `/foundation/brain.md` — Sections 3 and 6 updated on confirmation after relevant skills complete (Section 3 via `positioning-messaging` or `alternatives-map`; Section 6 via `proof-points`). `/context/skill-sessions.md` — one row per chained skill's own Learning Close, plus one row for the orchestration itself (Step 7). Master Program Document is produced in chat; ask the user where, if anywhere, to save it. - **Chat output format:** Program Charter → sequential skill outputs with checkpoints → coherence check results → Master Program Document. Each skill output is clearly delimited with skill name and date. - **External side effects:** None beyond confirmed brain writes. --- ## Verification Runs at program close, before Master Program Document is delivered. | Check | Standard | Pass = | |---|---|---| | Program Charter confirmed | User explicitly confirmed charter before any skill ran | Yes | | Each skill confirmed | Every output confirmed by user before next skill ran | Yes | | Coherence check run | /coherence run after all skills; conflicts surfaced and resolved | Yes | | Brain writes confirmed | Every brain write shown and confirmed before executing | Yes | | Master document complete | All required sections present with no orphaned outputs | Yes | | Next trigger surfaced | Next program cycle date recommended at close | Yes | --- ## Do Not Use For - **go-to-market-strategy** — when you only need a tier assignment and GTM brief, not a multi-skill program. Use that skill directly. - **positioning-messaging** — when you only need positioning or messaging work. Use that skill directly. - **retro** — for a single post-launch retrospective outside a larger program cycle. Use that skill directly. - **pre-mortem** — for standalone risk analysis before a launch. Use that skill directly, then bring the output into a full launch workflow if needed. --- ## Operating Rules - **A brain is required, a quick one is enough.** The orchestrator does not run without `/foundation/brain.md`; the Quick-Brain from `beachhead-segment` satisfies it. Disconnected skill outputs without shared context are worse than no program. - **Program Charter before execution.** No skill runs before the charter is confirmed. Starting without alignment is the most common reason programs produce wasted output. - **Checkpoint after every skill.** User confirms each output before the next skill runs. Silent chaining is not permitted — it produces coherence failures that are expensive to unwind. - **Coherence over speed.** If a conflict surfaces between outputs, stop and resolve it before proceeding. A fast program with incoherent outputs is not a program. - **Staleness must be surfaced.** If any brain section relevant to the workflow is stale, say so in the charter. The user decides whether to refresh — consciously. - **Brain writes require confirmation.** Never write to brain without explicit user go-ahead. Show what will be written, then wait. - **Skip requires downstream impact disclosure.** If the user skips a skill, tell them exactly what the next skill loses by not having that output. - **Read-only audits write nothing.** The Full PMM Onboarding / Audit workflow does not write to brain under any circumstance. - **One DRI per program.** If no DRI is named at intake, ask before confirming the charter. A program with no single owner has no owner. - **Next program trigger at close.** Every completed program ends with a recommended date for the next cycle. --- ## Quality Gate Runs at program close, before Master Program Document is delivered. | Check | Standard | Pass = | |---|---|---| | Program Charter confirmed | User explicitly confirmed charter before skills ran | Yes | | All skills accounted for | Each skill either ran or was explicitly skipped with reason | Yes | | Checkpoint for each skill | Each skill output confirmed before next skill ran | Yes | | Coherence check run | /coherence run after all skills — conflicts surfaced and resolved | Yes | | Brain writes confirmed | Every brain write shown to user and confirmed | Yes | | Master document complete | All required sections present in master document | Yes | | Open items documented | Unresolved gaps have named owner and due date | Yes | | Next trigger surfaced | Next program cycle date recommended at close | Yes | | No orphaned outputs | Every skill output referenced in master document | Yes | | DRI named | Program Charter includes a named DRI | Yes |
Comments (0)
Sign in to join the conversation.
Reviews (0)
No reviews yet.
No comments yet.