octavehq/abm

/octave:abm - Account Planner

Create account plans for target accounts by combining deep research, stakeholder mapping, persona matching, and an engagement sequence — all grounded in your library's Motion ICP cells and proof points. Output is a self-contained HTML account plan.

What this skill is. The unit of /octave:abm is the account. It produces a tight, account-level plan a rep acts on, structured as six jobs: (1) is this account worth it and why now, (2) what's likely driving them / what to explore, (3) who to engage (buying committee), (4) how to position per persona, (5) the engagement plan, (6) what could go wrong. It DOES advise — per-persona positioning and an entry-point engagement plan are the point — but at the account level. The distinction from /octave:meeting-prep is scope: meeting-prep coaches one scheduled meeting (talk tracks, discovery, room game-plan); abm plans the whole account pursuit across the committee. Single-asset copy still lives in /octave:generate.

Groundedness is non-negotiable. Every named person, title, employer, and LinkedIn URL in the buying committee must come from a real find_person/enrich_person result. Never synthesize a person or a LinkedIn slug. Honest gaps beat plausible inventions. A prior run shipped a fabricated committee — do not repeat it.

Principles

Follow these standards during generation. Read each before producing output.

Content and language:

• Editorial rules — no AI-isms, banned vocabulary, honest analyst tone
• Information principles — lead with conclusions, evidence-backed claims, narrative arc

Presentation:

• Presentation principles — universal visual rules, spacing, restraint
• HTML document format — scrollable-document visual rules

Octave data:

• Octave value — prioritize grounded workspace data over generic AI content

Usage

/octave:abm <company> [--stakeholders <N>] [--motion <name>] [--depth quick|full]

Examples

/octave:abm acme.com                                      # Full account plan
/octave:abm acme.com --stakeholders 5                     # Map top 5 stakeholders
/octave:abm acme.com --motion "Enterprise Expansion"      # Scope to a specific Motion's ICP matrix
/octave:abm acme.com --depth quick                        # Quick assessment only
/octave:abm "Acme Corp"                                   # Search by company name

Instructions

When the user runs /octave:abm:

Step 1: Identify Target Account

Parse input:

• Domain (e.g., acme.com) → Use directly
• Company name → Search for domain
• LinkedIn URL → Extract domain

If company name provided without domain:

find_company({ name: "<company_name>" })

Step 2: Deep Account Research

# Full company enrichment
enrich_company({ companyDomain: "<domain>" })

# Qualify against ICP
qualify_company({
  companyDomain: "<domain>",
  additionalContext: "Evaluate fit against all segments. Identify which segment, use cases, and Motion ICP cells are most relevant."
})

Step 3: Stakeholder Mapping (buying committee) — GROUNDEDNESS-CRITICAL

Build the committee persona by persona. For each relevant persona type in the library, use find_person to surface 1-2 strong-match real people — strong match only, real only. Do not pad to a headcount, and do not invent anyone: if a persona returns no strong match, leave that role as an honest gap.

# Persona types to search against
list_entities({ entityType: "persona" })

# For EACH relevant persona, find 1-2 strong matches at the account
find_person({
  searchMode: "people",
  companyDomain: "<domain>",
  fuzzyTitles: ["<titles for persona 1>"],
  limit: 3   // pick the 1-2 strongest from the results; do not list all
})
# Repeat per persona type (champion, economic buyer, evaluator, gatekeeper coverage)

Hard rule: the buying committee may only contain people find_person/enrich_person actually returned, with the LinkedIn URLs those tools returned. Never write a name, title, employer, or LinkedIn slug the tools did not give you. Tag anyone you could not confirm with .unconfirmed. Name empty roles as gaps ("no economic buyer identified yet").

For each key stakeholder found:

# Enrich top stakeholders
enrich_person({
  person: {
    firstName: "<first>",
    lastName: "<last>",
    companyDomain: "<domain>"
  }
})

# Qualify against personas
qualify_person({
  person: {
    firstName: "<first>",
    lastName: "<last>",
    companyDomain: "<domain>",
    jobTitle: "<title>"
  },
  additionalContext: "Match to our buyer personas. Identify their likely role in a buying decision (champion, evaluator, economic buyer, blocker)."
})

Step 4: Match Motion ICP Cell and Gather Intelligence

# Find the right Motion for this account (offering + motion type)
list_motions()

# See the persona × segment matrix for the matched Motion
list_motion_icps({ motionOId: "<motion_oId>" })

# Fetch the narrative for the target persona × segment cell
find_motion_icp({ motionIcpOId: "<motion_icp_oId>", includeLearnings: true })

# Find relevant proof points (industry, size match) — one per persona for Section 5
search_knowledge_base({
  query: "<company industry> <company size> results case study",
  entityTypes: ["proof_point", "reference"]
})

# Buying triggers for Section 3: use REAL library triggers if they match; otherwise
# frame pains as "likely / to explore" grounded in the motion cell — do not manufacture certainty
list_entities({ entityType: "buying_trigger" })

# Real engagement history for Section 6 (fold in if present; IGNORE dummy/seed pipeline values)
find_crm_activities({ entityType: "account", recordId: "<crm_account_id>" })   // if a real CRM account resolves
generate_crm_context({ companyDomain: "<domain>", objective: "account plan engagement history" })

# Check for competitive context
search_knowledge_base({
  query: "<company name> <any tech stack signals>",
  entityTypes: ["competitor"]
})

# Check for any existing conversation history
list_events({
  startDate: "<365 days ago>",
  filters: {
    companyDomains: ["<domain>"]
  }
})

Step 5: Resolve the Brand Kit

Resolve the brand before generating (do not skip this step). The document brand should be the workspace company's brand — that is, the Octave customer whose workspace you are operating in. The target account's logo appears in the masthead, but the document chrome (fonts, colors, header, footer) is workspace-company branded.

5a: Identify the workspace company. Call get_workspace_company to get the company name, domain/URL, and positioning.

5b: Resolve the workspace company's brand kit. Slugify the workspace company name and check for a cached brand kit at ~/.octave/brands/<slug>/manifest.json. If a complete kit exists (has manifest.json and tokens.css), use it automatically:

• inline the kit's tokens.css (:root + embedded @font-face) and ../get-brand-components/assets/kit_base.css into the output <style>;
• follow the kit's brand-kit.md → Signature moves; reuse the kit's real logo for header and footer, images/, and icons.json. If no complete kit exists → build one. Run the get-brand-components skill (read its SKILL.md and follow it) for the workspace company's domain. If the first attempt returns incomplete results, retry up to 3 times with different approaches (root domain, www. prefix, /about subpage). Only fall back to a clean default style after 3 genuine failures.

5c: Generic default is a last resort — only after the workspace company's brand kit cannot be built.

Step 6: Generate the Account Plan HTML

Load the shared rules and the document spec. Read each before producing output:

• Editorial rules — universal language rules, AI-ism kill list, banned vocabulary
• Information principles — lead with conclusions, evidence-backed claims, no interpretation padding (#7b applies hard here: information, not instructions)
• Presentation principles — universal visual rules, spacing, restraint
• HTML document format — scrollable-document visual rules
• Octave value — grounded workspace data over generic AI content
• Account plan structure — the six jobs, per-section specs, density rules, groundedness rule
• HTML scaffold — the locked CSS + section skeleton + persona-selector JS; reproduce this structure and card vocabulary
• Regression checklist — issues found during testing, verified against every new generation

Build a single, self-contained HTML file. No external dependencies beyond Google Fonts.

After writing the file, proceed immediately to Step 7 (Review Pipeline). Do NOT open the file in the browser or present it to the user yet.

Output Directory

.octave-abm/
└── <company-slug>-<YYYY-MM-DD>/
    └── account-plan.html

Write to the user's home directory (~/.octave-abm/...), not the skill or tuning folder.

Step 7: Review Pipeline — MANDATORY GATE

Do NOT open the plan in the browser, present the delivery summary, or tell the user the plan is ready until the review pipeline has completed and you have a scorecard.

Load the review protocol and execute the review loop against the generated HTML file. The protocol specifies the full process; here is the ABM-specific wiring:

7a: Mechanical lint (before spawning reviewers):

bash <skill-dir>/scripts/lint.sh <path-to-account-plan.html>

Fix every violation the lint surfaces.

7b: Spawn two reviewers in parallel (both Task calls in a single message):

Editorial reviewer:

Task tool:
  subagent_type: "octave-editorial-reviewer"
  prompt: "Review the file at [FILE PATH].
           Read these principle docs and run each Review Checklist:
           1. [skill-dir]/../shared/editorial-rules.md
           2. [skill-dir]/../shared/information-principles.md
           3. [skill-dir]/references/regression-checklist.md
              (Lean Intelligence section)
           Fix violations inline. Return scorecard."

Presentation reviewer:

Task tool:
  subagent_type: "octave-presentation-reviewer"
  prompt: "Review the file at [FILE PATH].
           Read these docs and run each Review Checklist:
           1. [skill-dir]/../shared/presentation-principles.md
           2. [skill-dir]/../shared/formats/html-document.md
           3. [skill-dir]/references/account-plan-template.md
           4. [skill-dir]/references/html-scaffold.md
              (structure + card vocabulary to reproduce)
           5. [skill-dir]/references/regression-checklist.md
              (Structure, Tightness, and Section 7 checks)
           Fix violations inline. Return scorecard."

7c: Loop decision. Read both scorecards:

| Cycle | 0 fixes | 1-2 fixes | 3+ fixes | |---|---|---|---| | Cycle 1 | CLEAN → 7d | Apply, loop | Apply, loop | | Cycle 2 | CLEAN → 7d | Apply, STOP | Apply, loop | | Cycle 3 (cap) | CLEAN → 7d | Apply, STOP | Apply, STOP |

Max 3 cycles. Re-run both reviewers each loop (back to 7b).

7d: Output combined scorecard to the user. This is proof the pipeline ran. Step 8 cannot start without it.

REVIEW PIPELINE COMPLETE
=========================
Editorial:      [N fixes / PASS]
Presentation:   [N fixes / PASS]

Total fixes: [N]
Cycles: [1-3]
Status: [CLEAN / N remaining issues]

Step 8: Deliver

After the review pipeline scorecard has been output:

1. Open the plan in the default browser.
2. Present a summary: file path, ICP score, stakeholders mapped (and missing roles), playbook matched, deal history found (Y/N), entry point.

Step 9: Sample outreach for the engagement plan (Section 6)

Generate the one or two sample emails that appear in the engagement plan (Section 6), for the entry-point persona. These are part of the document, not a post-delivery add-on — real and specific, grounded in the account's pains and the positioning from Section 5.

generate_email({
  person: {
    firstName: "<first>",
    lastName: "<last>",
    companyDomain: "<domain>",
    jobTitle: "<title>"
  },
  numEmails: 4,
  sequenceType: "COLD_OUTBOUND",
  allEmailsContext: "Account plan context: [company signals, persona match, Motion ICP narrative, proof points]",
  allEmailsInstructions: "Personalized to [company] specifically. Reference [relevant signals]. Use [proof points] progressively."
})

MCP Tools Used

Research

• enrich_company - Deep company intelligence
• enrich_person - Stakeholder background
• qualify_company - ICP fit scoring
• qualify_person - Persona matching
• find_person - Stakeholder discovery
• find_company - Company lookup by name

Library Context

• list_entities (persona) - Get persona definitions for stakeholder matching
• list_motions - List all Motions in the workspace
• list_motion_icps - List Motion ICP cells (persona × segment intersections) for a Motion
• find_motion_icp - Full Motion ICP cell narrative (Target ICP overview, Operating landscape, Strategic narrative, Pains and consequences, Benefits and impacts, Methodology, References) plus Learning Loop learnings
• search_knowledge_base - Proof points, references, competitive intel
• list_events - Existing conversation history with account
• get_workspace_company - Workspace company identity for brand resolution

Content Generation

• generate_email - Outreach sequences
• generate_content - Account-specific content
• generate_call_prep - Meeting preparation

Error Handling

Company Not Found:

Couldn't find "[input]".

Try:

1. Provide the company's website domain (e.g., acme.com)
2. Check spelling
3. Search by name: I'll look it up

No Stakeholders Found:

No contacts found at [Company] matching your personas.

Options:

1. Broaden the search (search all titles, not just persona matches)
2. Search for specific titles you know
3. Proceed with company-level plan only

Low ICP Score:

[Company] scored [X/100] against your ICP.

This is below typical qualification thresholds. Continue anyway? Or:

1. Find similar companies with better fit
2. See why the score is low and if any signals are missing
3. Proceed with adjusted expectations

Related Skills

• /octave:meeting-prep - Coached prep for a scheduled meeting (talk tracks, objection handling — the layer this skill deliberately omits)
• /octave:research - Deep-dive on a specific stakeholder
• /octave:ads - Ad campaign targeting this account
• /octave:battlecard-doc - Competitive intel if competitor detected
• /octave:pipeline - Coach on deal progression after engagement starts
• /octave:prospector - Find more accounts like this one