Grant Intake

Structured five-phase grant consultant intake to produce a Grant Readiness Summary and Organizational Profile.

When to Use

• User wants grant readiness or funder alignment
• User needs to capture org mission, programs, and logic models (inputs, activities, outputs, outcomes)
• User is preparing for grant applications and needs sector-specific metrics and a summary document

Workflow Summary

1. Phase 1 — Org name, mission, key programs
2. Phase 2 — Sector selection (Workforce Development, Health/Human Services, Green Economy, Education/Youth, Housing, Arts/Culture, Agriculture, or custom)
3. Phase 3 — Logic model per program (inputs, activities, outputs, outcomes)
4. Phase 4 — Sector-specific thematic questions with optional hints
5. Phase 5 — Review, confirm, and export Grant Readiness Summary + intake state

How to Run

• Seren Cloud agent: If the grant-intake cloud agent is available (e.g. via Seren Gateway or MCP), invoke it with a JSON body: { "state": null, "input": { "org_name": "...", "mission": "...", "key_programs": ["..."] } }. Response is { "state", "prompt", "outputs?" }; pass state back on each step to advance phases.
• Docs and deploy: See grant-intake deploy and invoke docs for endpoint URL, agent ID, and scripts.

Sectors

Predefined sectors: Workforce Development, Health/Human Services, Green Economy, Education/Youth, Housing, Arts/Culture, Agriculture. Custom sectors are supported.

Ralph Wiggum behavior (don't stop in the middle)

The skill is designed to keep going until the goal is reached and not drop information:

• Out-of-phase input: If the user provides information that belongs to another phase (e.g. sectors while in Phase 1), the skill stores it in the right place and acknowledges briefly ("I've noted your sectors for Phase 2."), then continues the current phase. Nothing is lost.
• Phase already filled: When entering a phase whose data was already provided earlier, the skill shows "Here's what we have: … Anything to add or change?" and waits for confirmation ("no" / "continue") or additional data before advancing.
• Unrecognized input: If the user says something that doesn't map to any phase, the skill asks one short clarifying question ("Did you mean to provide [expected fields], or something else?") and can store the reply as a note; it then continues the current step.
• Export with blanks: The user can confirm and export the summary even if some fields are still blank; the summary may have empty sections.
• Continue without changes: To move on when a phase already has data, the client can send continue_without_changes: true or no_changes: true in the input.