git:20260905.37629bc to git:20260906.f9cef19

55 added, 45 removed. Audit A to A.

---

name: deep_dive_invitation

- description: Create a review-only Markdown deep-dive invitation for a Dealum startup. Use when Joëlle provides a startup and founder or interested-investor contacts, including unstructured text copied from Dealum's Funding tab. Extract the relevant contacts from that text, invoke the Python workflow, and never send the email.

+ description: Create a review-only Markdown deep-dive invitation for Joëlle from a Dealum startup and supplied contacts. Use for invitation preparation; never send the email or create a Gmail draft.

---

# Deep-dive invitation

- Create a Markdown InsightFile for Joëlle to review. Never send the email and do

- not create a Gmail draft.

+ Prepare recipient reconciliation and an invitation for human review.

- ## Input preparation

+ ## Inputs and outputs

- Ask for the exact Dealum startup name or application code when it is not clear.

- The Python entry point accepts structured, comma-separated contacts:

+ The async `deep_dive_invitation(startup, founders=None, investors=None)` takes

+ lists of canonical `Person` objects and returns one `InsightFile` in a list,

+ named `deep-dive-invitation-<startup>-<model>.md`. `parse_people_csv` adapts

+ comma-separated names, emails or `Name <email>` entries.

- ```bash

- conda run -n sictic-env python -m skills.harness -- /deep_dive_invitation \

- --startup "<EXACT_DEALUM_NAME>" \

- --founders "Jane Founder <jane@example.com>, John Founder <john@example.com>" \

- --investors "Nina Member <nina@investor.sictic.ch>, FONGIT"

- ```

+ For pasted Dealum Funding-tab text, extract contacts from Internal and External

+ under FUNDING INTERESTS. Preserve names and emails, including name-only

+ organizations; ignore round amounts, tags and interface text. Do not invent

+ contacts. Ask for the exact startup name/code only when unclear.

- Both `--founders` and `--investors` are optional and plural. Each entry may be

- `Full Name <email>`, an email address, or a name. Use natural `First Last`

- names; commas separate people.

+ ## Workflow and dependencies

- When the user provides unstructured Dealum Funding-tab text, clean it before

- invoking the Python workflow:

+ Import Dealum data, request normal `startup_profile`, read member preferences,

+ then request `expert_search`. Existing member rosters and investor profiles are

+ required; the invitation does not discover people. It has no bulk registration.

- 1. Begin at `FUNDING INTERESTS`.

- 2. Extract contacts from both `Internal` and `External` sections.

- 3. Keep full names and email addresses. Preserve name-only contacts such as an

- organization, but never invent an email or LinkedIn ID.

- 4. Ignore round descriptions, amounts, tags, telephone numbers, dates,

- documents, discussions, and interface labels.

- 5. Convert the result to the structured `--investors` value above.

+ Supplied contacts are authoritative. Investor reconciliation tries exact email,

+ then LinkedIn ID, then normalized-name matches, not general fuzzy matching.

+ Investor email priority is the supplied address, then the matched member's

+ `@investor.sictic.ch`, `@sictic.ch`, then another stored address. Founder emails

+ can come from supplied contacts and the Dealum application contact; missing

+ addresses produce placeholders. Interested-investor Funding-tab extraction is

+ not automated, so the draft always includes a verification notice.

- Joëlle-supplied contacts, including contacts extracted from her pasted text,

- remain authoritative. The workflow augments them from Dealum and

- `persons_in_dataset("sictic-members")` without silently fuzzy-matching them.

+ Interested investors with an address remain in Cc even with preference `none`;

+ missing addresses create notices and are omitted until completed. Exclude interested

+ members and opted-out members from expert search. Configuration currently requests

+ 16 candidates and selects at most 10 usable experts. Bcc excludes addresses already

+ in To/Cc; experts appear only in Bcc, not as named rows in the email body.

- ## Workflow guarantees

+ Render the configured template mechanically after reconciliation. Cache lookup

+ occurs after dependencies; its key includes configuration, supplied people,

+ member preferences and expert-report content, with startup/community revisions.

+ Resolved member contacts are not independently hashed, so contact edits alone

+ may not invalidate the draft.

- - Dealum reconciliation must be reliable; Dealum import stops on no match or

- unresolved ambiguity.

- - Founder addresses may come from Joëlle and Dealum. Missing addresses create

- a visible placeholder and review notice.

- - Investor email priority is Joëlle, then Dealum when available, then a member

- `@investor.sictic.ch` or `@sictic.ch` address, then another member address.

- - Interested investors remain in Cc even when their invitation preference is

- `none`.

- - Expert search requests 16 candidates, excludes interested members and

- members whose `deep_dive_invitation` preference is `none`, and uses at most

- the first 10 usable experts.

- - Anyone in To or Cc is never included in Bcc.

- - Experts appear only as Bcc recipients, never as named rows in the email body.

- - The draft begins with warnings and reconciliation details, followed by the

- email. The body template lives in `config/deep_dive_invitation`.

- - Automatic Funding-tab extraction is not configured, so every draft includes

- a prominent verification notice.

+ ## Side effects and failure behavior

+ Dependencies may import, synchronize, generate profiles and rank experts. Save

+ only a review artifact: do not send messages, create mail drafts or change stages.

+ Missing/ambiguous contacts can become visible review notices; dependency and

+ generation failures propagate. Preserve the configured recipient precedence and

+ manual output selection.

+ ## Usage

+ ```bash

+ conda run -n sictic-env python -m skills.harness '/deep_dive_invitation --startup "<EXACT_NAME_OR_CODE>" --founders "Jane Founder <jane@example.com>" --investors "Nina Member <nina@example.com>"'

+ ```

+ The direct CLI uses the same plural `--founders` and `--investors` options;

+ both are optional.

+ ## References

+ - [Implementation and recipient precedence](deep_dive_invitation.py)

+ - [Template and settings](../../config/deep_dive_invitation/)

+ - [Member preferences](../member_preferences/SKILL.md)

+ - [Expert search](../expert_search/SKILL.md)