Founder Sprint

Founder Sprint (Phase 2)

Child of founder. Lean gatekeeper and state machine managing the founder journey from idea to forge handoff. Four stages with gate enforcement. Picks the next subskill. Updates venture-brief. NOT a planner, NOT an executor -- a gatekeeper.

Scope: Stage management and gate enforcement only. Sprint picks the right subskill to invoke, checks gate criteria, and advances the venture-brief state. It does NOT do task decomposition, work package creation, owner assignment, or sequencing beyond "which subskill next." Task decomposition is bob's job after forge takes over.

Siblings (parent = founder):

• founder-ideation — Phase 1 — adversarial brainstorm + data grounding
• founder-validation — Phase 2 — experiments, interviews, evidence capture
• founder-business-model — Phase 2 — calculator mode unit economics
• founder-gtm — Phase 3 (deferred) — positioning, distribution, channel selection

Inherited hard rules (from parent founder): HR-1 through HR-11 all apply. Additionally, HR-V1 through HR-V5 (from validation) and HR-BM1 through HR-BM6 (from business-model) are enforced through the subskill invocations sprint orchestrates.

The Four Stages

Stage 1: DIAGNOSE

Question: Is there anything worth validating?

Invokes: founder-ideation (Phase 1)

Gate criteria:

• venture-brief has >= 3 ranked ideas with data citations (HR-5)
• 1 idea selected (status: candidate or better, marked as selected)

• = 3 assumptions listed in venture-brief.assumptions[]

• = 3 kill criteria across the selected idea (HR-4)

Venture-brief updates on stage entry:

sprint_state:
  stage: diagnose

Venture-brief updates on gate pass:

sprint_state:
  stage_completed_at:
    diagnose: <timestamp>

Abort condition: All ideas scored below kill threshold by adversarial cross-fire.

sprint_state:
  stage: aborted
  outcome: "no viable ideas — all scored below kill threshold"
  abort_reason: "adversarial cross-fire killed all candidate ideas"

Stage 2: EVIDENCE

Question: Do we have enough evidence to commit?

Invokes: founder-validation (Phase 2)

Gate criteria:

• Top-3 riskiest assumptions each have >= 1 experiment with recorded evidence

• = 1 real interview logged (interview_count >= 1, HR-V2)

• No high-risk viability assumption falsified without either:
  • A documented pivot (pivots[] entry), OR
  • An explicit accepted_risk disposition with user-stated reasoning

Venture-brief updates on stage entry:

sprint_state:
  stage: evidence

Venture-brief updates on gate pass:

sprint_state:
  stage_completed_at:
    evidence: <timestamp>
interview_count: <N>

Abort condition: Viability assumption falsified, user refuses pivot or accept-risk.

sprint_state:
  stage: aborted
  outcome: "killed at evidence gate"
  abort_reason: "viability assumption '[claim]' falsified; user declined pivot"

Stage 3: DECISION

Question: Is this a viable business we should build?

Invokes: founder-business-model (Phase 2), optionally conversion-psychology for CTA work

Gate criteria:

• business_model block populated in venture-brief:
  • price with tag
  • unit_econ ranges (contribution_margin at minimum)
  • decision_rule stated
  • decision_verdict: GREEN or CONDITIONAL_GO with user acknowledgment
• forge_brief block drafted:
  • problem (non-empty)
  • solution (non-empty)
  • success_criteria (non-empty list)
  • non_goals (non-empty list)

Venture-brief updates on stage entry:

sprint_state:
  stage: decision

Venture-brief updates on gate pass:

sprint_state:
  stage_completed_at:
    decision: <timestamp>
business_model: { ... }      # populated by founder-business-model
forge_brief: { ... }         # drafted during this stage

Abort condition: Calculator mode returns RED on decision rule AND user declines to adjust.

sprint_state:
  stage: aborted
  outcome: "killed at decision gate — unit economics"
  abort_reason: "decision rule returned RED; user declined to adjust pricing/costs"

Stage 4: HANDOFF

Question: Are we ready to build?

Invokes: Nothing — runs a checklist, flips the flag, invokes forge.

Gate criteria:

• All previous stages have stage_completed_at timestamps
• forge_brief populated with all required fields
• User gives explicit "ship it" confirmation

Venture-brief updates on gate pass:

sprint_state:
  stage: handoff
  stage_completed_at:
    handoff: <timestamp>
forge_handoff_ready: true
handoff_at: <timestamp>

Handoff method: Sprint routes to forge with explicit spawn prompt:

came_from_founder: true
venture_brief_path: .founder/venture-brief.yaml
forge_brief:
  problem: <from venture-brief>
  solution: <from venture-brief>
  success_criteria: <from venture-brief>
  non_goals: <from venture-brief>
  complexity_hint: <from venture-brief>
  open_questions: <from venture-brief>

Forge reads the brief from the spawn prompt, NOT from ambient session-start discovery.

Abort condition: User changes mind.

sprint_state:
  stage: handoff
  outcome: "paused at handoff"
forge_handoff_ready: false
# forge_brief remains intact for later

Sprint Operation

On Entry

1. Read venture-brief.yaml. Refuse if missing.
2. Check schema_version. Must be 2. Refuse with migration guidance if 1.
3. Read sprint_state.stage to determine current position.
4. If no sprint_state exists, initialize at diagnose.
5. Present current state to user:

   "Your venture is at stage [STAGE]. Here's what we need to advance: [list gate criteria with checked/unchecked status]"

Advancing

1. Check ALL gate criteria for current stage.
2. If all pass: advance to next stage. Update venture-brief.
3. If some fail: show what's missing and invoke the appropriate subskill.
4. Never auto-advance through multiple stages in one invocation.

Subskill Picker

Resetting

User can request reset to an earlier stage:

• Reset clears the later stage_completed_at timestamps
• Reset does NOT delete evidence, experiments, or business model data
• Reset records a pivots[] entry explaining why
• Reset is an explicit user action, not automatic

Cross-Stage Rules

1. Brief is authoritative — when sprint and subskills disagree, venture-brief.yaml wins
2. No stage-skipping — sprint refuses to jump. Reset allowed.
3. Evidence or no transition — each gate requires artifacts, not assertions
4. Aborts are recorded — outcome + reason written to brief so next invocation knows
5. Sprint does NOT do task decomposition — that's bob's job after forge takes over
6. If no venture-brief.yaml exists — refuse; tell user to run founder-ideation first

Venture-Brief Integration

Sprint reads and writes the following venture-brief fields:

See references/venture-brief-state-rules.md for the complete field-level contract.

Failure Modes

Anti-Patterns

Reference Files

Read these as needed during sprint operation:

• references/stage-machine-spec.md — complete state transition diagram, allowed/forbidden paths, edge cases
• references/gate-criteria.md — detailed checklist per gate with specific evidence artifacts required and pass/fail rubrics
• references/venture-brief-state-rules.md — which fields sprint reads/writes per stage, schema version check, migration notes
• references/forge-handoff-protocol.md — explicit handoff spawn prompt template, came_from_founder flag, what forge does on receipt

When NOT to Use This Skill

• User wants quick ideas without committing to a sprint — use founder-ideation directly
• User wants to validate without the stage machine — use founder-validation directly
• User wants unit economics without the stage machine — use founder-business-model directly
• User is already building (post-forge handoff) — sprint is done; use forge/bob directly
• User wants legal/tax/valuation advice — REFUSED (HR-1, HR-2)
• Sprint is optional — users CAN invoke subskills directly without going through sprint