landscape-scan

Files (3)

• SKILL.md
• landscape-schema.md · references/
• query-templates.md · references/

SKILL.md (first lines)

# /landscape-scan -- Ground-Truth Ecosystem Snapshot

Every claim your marketing makes sits on top of an assumption about the market.
"We're the first to..." "Nobody else does..." "The market is moving toward..."
When these assumptions are wrong, your positioning collapses on contact with
reality. Customers who Google your claims and find them false don't come back.

This skill builds a verified ecosystem snapshot that becomes the single source
of truth for all downstream content. The Claims Blacklist it produces hard-gates
every content skill -- if a claim is blacklisted, no skill writes it. Period.

No SaaS tools needed. Live web research + user validation.

---

## On Activation

1. Check if `brand/` directory exists in the project root.
2. If it does, read available files in priority order:
   - `competitors.md` -- existing competitive intel (primary input)
   - `positioning.md` -- current positioning angles and claims
   - `audience.md` -- target market and buyer personas
   - `voice-profile.md` -- brand personality (for tone of output)
   - `learnings.md` -- past marketing learnings and corrections
3. Apply loaded brand context to focus the landscape scan -- existing competitor
   data narrows the research query, positioning data identifies claims to verify.
4. If `brand/` does not exist or is empty, proceed without it -- this skill
   works standalone by asking the user foundational questions.

---

## Iteration Detection

Before starting, check whether `./brand/landscape.md` already exists.

### If landscape.md EXISTS --> Refresh Mode

Do not start from scratch. Instead:

1. Read the existing landscape file.
2. Present a summary of the current state:
   ```
   EXISTING LANDSCAPE SNAPSHOT
   Last updated {date} by /landscape-scan

   Ecosystem segments: {N}
   Claims blacklisted: {N}
   Market shifts tracked: {N}

   Freshness: {days} days old (threshold: 14 days)

   ------------------------------------------

   What would you like to do?

   1. Full refresh -- re-run /last30days, revalidate everything
   2. Verify claims -- check if blacklisted claims are still invalid
   3. Add new segment -- expand to cover a new market area
   4. Rebuild from scratch -- discard and start over
   ```

3. Process the user's choice:
   - Option 1 --> Re-research with fresh /last30days query, merge with existing data
   - Option 2 --> Focused validation of Claims Blacklist items only
   - Option 3 --> Targeted research on a new ecosystem segment, merge into existing file
   - Option 4 --> Full process from scratch

4. Before overwriting, show a diff of what changed and ask for confirmation.

### If landscape.md DOES NOT EXIST --> Full Scan Mode

Proceed to the full process below.

---

## The Core Job

Map current ecosystem reality so every downstream skill operates on verified
ground truth. The output has two critical components:

1. **Ecosystem Snapshot** -- what is actually happening in the market right now
   (players, trends, shifts, emerging categories, consolidation, funding rounds)
2. **Claims Blacklist** -- specific claims your marketing MUST NOT make because
   they are verifiably false, outdated, or easily disproven by a customer who
   spends 30 seconds searching

The Claims Blacklist is not advisory. It is a hard gate. Every content-generating
skill reads `brand/landscape.md` and refuses to write blacklisted claims.

---

## Orchestration Flow

### Phase 1: Preflight

1. Read brand/ context files (see On Activation above).
2. Detect mode: create vs refresh (see Iteration Detection above).
3. Determine context level:
   - **L0** -- No brand/ files. Need to ask everything.
   - **L1** -- Have product name/description only.
   - **L2** -- Have competitors.md and/or positioning.md.
   - **L3** -- Have L2 + will run /last30days for live data.
   - **L4** -- Have L3 + existing landscape.md (refresh mode).
4. If `--dry-run` is set, report what WOULD happen and exit. Do not make any
   network calls or write any files.

### Phase 2: Build Research Query

Extract search parameters from available context:

- **Product/Category**: from positioning.md `primary_gap` or user input
- **Competitors**: from competitors.md competitor names, or user input
- **Target market**: from audience.md `primary_persona`, or user input
- **Time window**: last 30 days (default), or user-specified

Build the research query (see `references/query-templates.md` for industry-specific
templates). The query should cover:

1. Market movements -- funding, acquisitions, pivots, shutdowns

From MoizIbnYousaf/marketing-cli, MIT.

More from MoizIbnYousaf/marketing-cli