pitch

<purpose> Main workflow for heres-the-thing. Guided interview that builds a campaign: gathers subject, audience, goals, context, and precedent, then generates a strategy brief and deliverables based on the decision framework.

If --resume or an existing campaign-id is passed, loads the existing campaign and skips to prep level assessment (delegates to /heres-the-thing:prep logic). </purpose>

<steps> <step id="load-config" number="1"> <description>Load Configuration</description>

<load-config>
Resolve the user's home directory (run `echo $HOME` via Bash). Use this absolute path for all file operations below -- never pass `~` to the Read tool.

1. Read `<home>/.things/config.json`
   <if condition="config-missing">Tell the user: "Run `/things:setup-things` first." Then stop.</if>

2. Check that `<home>/.things/heres-the-thing/` exists
   <if condition="htt-missing">Tell the user: "Run `/heres-the-thing:setup-htt` first." Then stop.</if>

3. Read `<home>/.things/heres-the-thing/deliverable-types/index.json`

4. Read `<home>/.things/shared/professional-profile.json` (for self-assessment defaults)

5. <if condition="resume-flag-or-campaign-id">
   Read the existing campaign.json and skip to Step 8 (Generate).
   </if>
</load-config>

</step> <step id="gather-subject" number="2"> <description>Step 1: What Are You Positioning?</description>

Use AskUserQuestion:

What are you positioning? This can be anything -- a tool you built, someone else's idea, a proposal, a promotion, a policy change, a request for resources, or something that doesn't exist yet.

<phase name="subject-source" number="1">
How do you want to describe it?
<options>
- Describe it now (free text)
- Pull from an i-did-a-thing log
- Reference a URL, file, or existing campaign
- Describe something that doesn't exist yet (idea, aspiration)
</options>

<if condition="pull-from-idat">
Search `<home>/.things/i-did-a-thing/index.json` for recent entries. Present top 10 by date. Let user select one.
Extract title, description, tags, and store as `source_refs` entry with type `idat_log`.
</if>

<if condition="reference-url">
Ask for the URL or file path. Store as `source_refs` entry with type `url`.
</if>
</phase>

<phase name="subject-type" number="2">
What kind of thing is this?

Use AskUserQuestion (free text accepted):
<options>
- project -- Something you built or are building
- idea -- A concept or proposal
- request -- Asking for resources, approval, or support
- tool -- A specific tool, library, or system
</options>

<rule>Accept any string -- these are suggestions, not an enum.</rule>
</phase>

<phase name="subject-familiarity" number="3">
How well do you understand what you'll be communicating about?

Use AskUserQuestion:
<options>
- expert -- I know this deeply, could answer any question
- i_built_it -- I created it and know it well
- familiar -- I understand the basics but may have gaps
- new_to_me -- I'm still learning about this
</options>

<if condition="new_to_me-or-familiar">
Flag: Subject research tasks will be included in the strategy brief. Confidence flags will mark assumptions that need verification.
</if>
</phase>

</step> <step id="gather-goals" number="3"> <description>Step 2: Who / Why / Where / When (Per Goal)</description>

<constraint>A campaign can have multiple goals. Start with the first one, then ask if they want to add more.</constraint>

For each goal, use AskUserQuestion to gather:

<phase name="goal-basics" number="1">
1. **Goal description**: What outcome are you trying to achieve?
2. **Goal type**:
   <options>
   - short_term -- A specific near-term outcome
   - long_term -- A broader objective that may require multiple steps
   </options>
3. **Target date**: When does this need to happen? (date or relative like "next Friday", "end of Q2")
4. **Stakes**:
   <options>
   - low -- Nice to have, won't hurt if it doesn't land
   - medium -- Important but recoverable
   - high -- Critical outcome, significant consequences
   </options>
</phase>

<phase name="goal-audience" number="2">
**Audience**: Who are you trying to reach?

Use AskUserQuestion:
<options>
- A specific person (I'll name them)
- An audience segment I've already defined
- A new audience I'll describe now
</options>

<if condition="specific-person">
Ask for their name. Check `<home>/.things/shared/people/` for an existing profile.
<if condition="person-exists">Load their profile.</if>
<if condition="person-not-found">
Ask the user: "No profile found for <name>. I'll create one. What's their role/title?"
Create a minimal profile in `<home>/.things/shared/people/<slug>/profile.md` and `index.json`.
</if>

Then ask:
- **Receptiveness**: How receptive is this person to what you're proposing?
  <options>
  - champion -- They're already on your side
  - enthusiastic -- Open and supportive
  - neutral -- Will evaluate on merits
  - skeptical -- Needs convincing
  - hostile -- Actively against it
  </options>
- **Familiarity**: How well do you know this person?
  <options>
  - well_known -- Worked together extensively
  - moderate -- Some interaction
  - low -- Barely know them
  - unknown -- Never met
  </options>
- **Familiarity notes** (free text, optional): Anything else about your relationship or their preferences?
</if>

<if condition="existing-segment">
List available segments from `<home>/.things/heres-the-thing/audiences/`. Let user select.
Ask for receptiveness override (or use baseline).
</if>

<if condition="new-audience">
Gather basic audience info inline (name, key concerns, decision criteria). Store in campaign only (not as a reusable segment).
</if>

<rule>A goal can have multiple audience entries (e.g., presenting to a group).</rule>
Ask: "Is there anyone else in this audience for this goal?"
</phase>

<phase name="goal-medium" number="3">
**Medium**: How will you deliver this?

Use AskUserQuestion (free text accepted):
<options>
- meeting_1on1 -- One-on-one meeting
- meeting_group -- Group meeting or presentation
- email -- Written email
- proposal -- Formal written proposal
- talk -- Presentation or talk
- cold_outreach -- Cold email or message
- slack -- Slack message or thread
</options>

<rule>Accept any string -- these are suggestions, not an enum.</rule>

**Delivery length** (if applicable): How long do you have? (e.g., "30 min", "1 page", "5 min lightning talk")
</phase>

<phase name="goal-deliverables" number="4">
Based on the decision framework (see Step 7), the system will auto-select deliverables. But ask:

Are there specific deliverables you want for this goal beyond the auto-selected ones?

Show available types from the deliverable type registry.
</phase>

<phase name="more-goals" number="5">
Use AskUserQuestion:
Do you want to add another goal for this campaign?
<options>
- Yes -- add another goal
- No -- that's all
</options>

<if condition="add-another">Loop back to the start of this step.</if>

<rule>If a user defines both short-term and long-term goals, prompt them to link them: "Does <short-term-goal> roll up to <long-term-goal>?" Set `parent_goal` accordingly.</rule>
</phase>

</step> <step id="gather-context" number="4"> <description>Step 3: Context</description>

Use AskUserQuestion:

1. **Situational context**: What's the landscape? What's happening that makes this timely? (free text)

2. **Precedent**: Is there anything relevant from your past that informs this?
   <options>
   - Yes -- a previous campaign
   - Yes -- an i-did-a-thing log entry
   - Yes -- a URL, PR comment, or document
   - Yes -- multiple things (I'll list them)
   - No precedent
   </options>

   <if condition="has-precedent">
   For each precedent item:
   - Ask for the type and reference
   - Ask for a relevance note: "Why does this matter for this campaign?"
   Store as structured `precedent` array entries.
   </if>

</step> <step id="self-assessment" number="5"> <description>Step 4: Self-Assessment</description>

Read `<home>/.things/shared/professional-profile.json`.

<if condition="profile-has-strengths">
Show the user their known strengths and ask:
"These are the communication strengths I know about. Any to add or highlight for this campaign?"
</if>

<if condition="profile-missing-or-empty">
Use AskUserQuestion:
- What are you good at when communicating? (e.g., concrete examples, data-driven arguments, storytelling)
- What do you tend to struggle with? (e.g., handling curveball questions, staying concise, reading the room)
</if>

</step> <step id="auto-tags" number="6"> <description>Step 5: Tags</description>

Auto-suggest tags from:
- Subject type and description keywords
- Goal descriptions
- Audience names/segments
- Medium types

Present suggested tags and ask user to confirm/edit/add.

</step> <step id="write-campaign" number="7"> <description>Write Campaign</description>

Generate a campaign ID from the subject title (lowercase, hyphens).

```bash
mkdir -p <home>/.things/heres-the-thing/campaigns/<campaign-id>/strategy
mkdir -p <home>/.things/heres-the-thing/campaigns/<campaign-id>/artifacts
mkdir -p <home>/.things/heres-the-thing/campaigns/<campaign-id>/outcomes
```

<output-path>`<home>/.things/heres-the-thing/campaigns/<campaign-id>/campaign.json`</output-path>

Write the full campaign JSON following the schema from the design:

```json
{
  "id": "<campaign-id>",
  "status": "active",
  "created": "<current_date>",
  "tags": ["<tags>"],
  "subject": {
    "title": "<title>",
    "type": "<type>",
    "familiarity": "<familiarity>",
    "description": "<description>",
    "source_refs": [<refs>]
  },
  "context": "<context>",
  "precedent": [<precedent_entries>],
  "goals": [
    {
      "id": "<goal-id>",
      "type": "<short_term|long_term>",
      "description": "<description>",
      "parent_goal": "<parent-id or null>",
      "target_date": "<date>",
      "status": "active",
      "stakes": "<low|medium|high>",
      "tags": ["<goal-tags>"],
      "audience": [<audience_entries>],
      "medium": "<medium>",
      "delivery_length": "<length or null>",
      "deliverables": ["<type_ids>"]
    }
  ]
}
```

</step> <step id="generate-deliverables" number="8"> <description>Step 6: Generate Strategy Brief + Deliverables</description>

For each goal in the campaign:

<phase name="time-assessment" number="1">
Calculate time available from `target_date` minus current date:
- ≤15 min or same day: **urgent**
- Hours to end of day: **short**
- Days: **moderate**
- Weeks+: **extended**
</phase>

<phase name="strategy-brief" number="2">
**Always generate a strategy brief.**

<output-path>`<home>/.things/heres-the-thing/campaigns/<campaign-id>/strategy/<goal-id>-<timestamp>.md`</output-path>

Write the full 5-section strategy brief:

```markdown
# Strategy Brief: <goal-id>
## Campaign: <campaign-id>
## Version: <timestamp>

---

## 1. Situation Analysis
[Current landscape, timing, why now, relevant precedent from campaign.precedent]

## 2. Audience Profile
[Who they are (from audience data), what they care about, known objections, decision criteria, receptiveness assessment]
[If audience familiarity is low/unknown: include research tasks -- "Find out X about Y before proceeding"]

## 3. Subject Positioning
[What you're pitching, key value props framed for THIS audience, differentiators vs. alternatives]
[If subject.familiarity is new_to_me or familiar: add confidence flags -- "This framing assumes X -- verify before delivery"]
[If audience is champion: frame as ammo they can use to advocate on your behalf]

## 4. Communication Strategy
[Medium rationale, framing approach, opening hook, key messages in priority order]
[Language to use, language to avoid]
[If audience is skeptical/hostile: defensive framing, objection-first approach]

## 5. Execution Plan
[Prep timeline based on time available, delivery approach, contingencies, follow-up plan, success metrics]

---

### Appendix A: Source Materials
[Links to IDAT logs, URLs, reference docs from subject.source_refs]

### Appendix B: Precedent
[Referenced precedent with relevance notes and lessons applied]
```

<rule>Time modulates depth, not type. Urgent = concise sections. Extended = deep research and multiple iterations.</rule>
</phase>

<phase name="decision-tree" number="3">
**Walk the additive decision tree for each goal.**

Check each branch independently:

1. **Medium branch**:
   - Sync medium (meeting, talk, presentation) → generate `meeting_prep_doc`
   - Async medium (email, proposal, document) → generate `artifact_draft` + any custom types in `deliverables`

2. **Stakes branch**:
   - stakes >= high → generate `objection_map`, offer Q&A prep session

3. **Audience receptiveness branch**:
   - skeptical/hostile → generate `objection_map` (if not already), reframe strategy defensively
   - champion → add "equip mode" section to strategy brief (ammo they can use to advocate)

4. **Subject familiarity branch**:
   - new_to_me or familiar → add subject research tasks and confidence flags to brief

5. **Audience familiarity branch**:
   - unknown or low → add audience research tasks to brief

6. **Time branch**:
   - Days+ to prepare → generate `prep_plan`
   - If extended time + high stakes → offer rehearsal plan
</phase>

<phase name="medium-modifiers" number="4">
**Apply medium-specific additions:**

| Medium contains | Additional content |
|---|---|
| `talk` or `presentation` | Add slide outline, timing breakdown, transition phrases to artifacts |
| `proposal` | Add executive summary section, decision framework for reader |
| `cold_outreach` or `cold_email` | Add subject line options, follow-up sequence (day 3, day 7) |
| `meeting_group` | Add role-specific talking points per audience member |
</phase>

<phase name="write-deliverables" number="5">
For each deliverable determined by the tree walk:

<if condition="builtin-type">
Generate the deliverable directly based on the strategy brief and campaign context.
Write to `<home>/.things/heres-the-thing/campaigns/<campaign-id>/artifacts/<goal-id>-<type-id>-<timestamp>.md`
</if>

<if condition="custom-type">
Use the Task tool to launch the deliverable-agent subagent with:
- campaign_path
- goal_id
- type_id
- strategy_brief_path
</if>
</phase>

<phase name="always-offer" number="6">
After generating deliverables, always offer:
- `follow_up_template` (if the goal has next_steps potential)
- Any registered custom deliverable types not already generated
- Q&A prep session (via `/heres-the-thing:prep`)
</phase>

</step> <step id="summary" number="9"> <description>Summary</description>

<completion-message>
Campaign created: `<campaign-id>`

Subject: <title>
Status: active
Tags: <tags>

Goals:
<for each goal>
  <goal-id> (<type>) -- <description>
    Target: <date> (<time_available>)
    Audience: <audience summary>
    Medium: <medium>
    Stakes: <stakes>
    Deliverables generated:
      - strategy/<goal-id>-<timestamp>.md (strategy brief)
      - artifacts/<goal-id>-<type>-<timestamp>.md (for each)

</for>

Next steps:
- `/heres-the-thing:prep <campaign-id>` -- refine deliverables or run Q&A prep
- `/heres-the-thing:outcome <campaign-id>` -- log what happened after delivery
- `/heres-the-thing:review` -- analyze patterns across campaigns
</completion-message>

</step> </steps>