Skip to content
Agentmart
Submit a tool

strategy-pipeline

Phase 11 procedure — capabilities → set_strategy_inputs → plan → apply → approve, plus 3-path blocker resolution (ask / build / defer).

Skill by petr-royce

IndexedMIT★ 106
Write a review

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/elophanto-strategy-pipeline

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Phase 11 procedure — capabilities → set_strategy_inputs → plan → apply → approve, plus 3-path blocker resolution (ask / build / defer). Covers WHOLE business surface via injected OPERATIONAL CONTEXT, not just what_we_sell.

Use cases: data

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/strategy-pipeline

What's inside

Skill name
strategy-pipeline
Description
Phase 11 procedure — capabilities → set_strategy_inputs → plan → apply → approve, plus 3-path blocker resolution (ask / build / defer). Covers WHOLE business surface via injected OPERATIONAL CONTEXT, not just what_we_sell.
License
MIT
Metadata
phase: 11 · author: petr-royce · version: 1.1.0
Instructions
95 lines in SKILL.md · view on GitHub

Files (1)

  • SKILL.md

SKILL.md (first lines)

# Strategy pipeline (Phase 11)

Called by [[drive-business]] PATH B, or directly when the operator asks to plan an existing productized company.

## Triggers

- company_plan
- company_capabilities
- company_plan_apply
- set up strategy
- capability audit
- 30 day plan
- marketing plan
- blockers
- strategy inputs

## Pipeline (5 tools, in order)

### 1. `company_capabilities(<slug>)`
Audit vault + tools + skills. Writes `capabilities.md`. Returns structured map.

### 2. Research, propose recommendations, then capture inputs
Do NOT ask the operator blank questions. The agent has the research tools to ground every field. Operator's job is approve / modify / override, not fill blanks.

**Research first** (parallel where possible):
- `browser_navigate(<url>)` + `browser_extract` (homepage, /about, /pricing, /hire, /faq)
- `web_search('<domain> competitors')` and `web_search('<industry> alternatives 2026')`
- `company_report(<slug>)` — ledger gaps, pipeline state
- `file_read companies/<slug>/company.yaml` — declared channels, KPIs

**Propose each field with one-line rationale** in chat as a table:

| Field | Recommended | Rationale |
|---|---|---|
| `target_audience` | <derived from homepage + /hire positioning> | <evidence> |
| `competitors` | <3-5 from web_search> | <industry/category> |
| `current_challenges` | <from ledger gaps + funnel friction> | <evidence> |
| `unique_selling_points` | <from homepage value props> | <what only this does> |
| `primary_goals` | <3-5 from list below> | <inferred from product type> |
| `strategy_mode` | <one of: standard / unconventional / guerrilla / brand-awareness / controversial> | <from brand tone> |
| `focus` | <full / seo / geo / content / paid / social / email / brand> | <from active channels> |
| `budget_type` + `_amount` + `_period` | suggest `organic / 0 / monthly` if no signal | operator-decided |
| `risk_tolerance` | suggest 50 (balanced) unless brand tone suggests higher | operator-decided |
| `timeline_hint` | suggest based on stage (e.g. "first paid customer in 30d") | operator-decided |

Valid `primary_goals`: Brand Awareness, Lead Gen, Sales Growth, Retention, Thought Leadership, Community Building, Content Authority, Market Disruption, Competitive Advantage, AI Search Visibility.

Add: *"Approve all, or tell me which to change."*

**After operator confirms/modifies**: `company_set_strategy_inputs(slug, …)` — MODERATE. Uses the operator-approved values.

### 3. `company_plan(<slug>, override_strategy_mode=…, override_focus=…)`
Deterministically prepends OPERATIONAL CONTEXT (active schedules + last-7d ledger + prospect funnel + missions) so the strategy covers EVERY surface, not just `what_we_sell`. Writes `strategy/proposed/<timestamp>.yaml`.

### 4. `company_plan_apply(<slug>)` — MODERATE
Atomically:
- Detects blockers (5 types — see below)
- Mission + N goals (tactic_meta packed in plan_json) + schedules (from `timeline.month1/2/3` via `execution_priority`)
- Writes `voice_proposed.yaml` from `creativeDirections.hookTemplates`
- Writes `blockers.yaml`
- Promotes proposed → active, archives prior

### 5. `company_plan_approve(<slug>, note="…")` — MODERATE
Operator finalize. Touches the strategy mission so arbiter ranks high.

## Blocker resolution (3 paths)

Each blocker in `blockers.yaml` has `resolution_proposal`:

- **`ask`**: operator must act (credentials, business decisions). Surface in chat with `build_hint`.
- **`build`**: `from_buildable_blockers` arbiter candidate invokes `self_create_plugin` (`build_method`) or `skill_promote`. CRITICAL permission gates each build. In chat mode, only attempt if operator explicitly OKs.
- **`defer`**: operator marked tactic out of scope. Note and move on.

## Surface coverage (CRITICAL)

The OPERATIONAL CONTEXT block lists every active surface (X growth, Pump.fun, email, prospect funnel, etc.). The resulting strategy MUST address each — not just `what_we_sell`. A single-surface strategy for a multi-surface business is a quality failure.

## Hard rules

- ❌ Don't skip `company_capabilities` — no operational context = narrow strategy.
- ❌ Don't skip operator approval of `strategy_inputs` — those calls aren't yours.
- ❌ Don't promote proposal to active via file_write — use `company_plan_apply`.
- ❌ Don't call `apply` + `approve` back-to-back without surfacing blockers between.
- ❌ Don't invoke `self_create_plugin` on a build blocker without explicit chat OK.
- ❌ Don't re-plan a company with an active strategy unless operator asks.

## Verify

- [ ] 5 steps ran in order
- [ ] `strategy/active/strategy.yaml` exists
- [ ] `blockers.yaml` surfaced to operator
- [ ] Voice seeded if hookTemplates populated
- [ ] Tactics span all surfaces from operational context

From elophanto/EloPhanto, MIT.

More from elophanto/EloPhanto

Signals

Stars
106
Forks
9
Open issues
2
Last commit (this folder)
May 27, 2026 (5 months ago)
Last release
Oct 1, 2026 (29 days ago)

From GitHub (fetched Oct 24, 2026).

Details

Version
1.1.0
License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from petr-royce

Cite this page

strategy-pipeline skill. Agent Mart. https://agentmart.dev/skills/elophanto-strategy-pipeline (last updated 2026-10-24)