Keysat
536358093f
The intake bot now accepts a photo of a business card in the intake room and turns it into the same new-investor proposal a typed note would. The only new step is image -> text; everything downstream (parse, fuzzy match, in-thread approval, log-communication write) is reused unchanged. M3 was deferred only because Spark Control had no vision model. That blocker is gone: the daily-driver Qwen is vision-capable under the same model id, and the gateway forwards OpenAI multimodal content untouched, so no gateway/server/s9pk change is needed -- this ships bot-only (git pull + rebuild on the Spark). Transcribe-then-reuse (not vision-straight-to-JSON) is deliberate: the transcription becomes the source text the email-integrity rule checks against, so a mis-read address can't reach the CRM unapproved -- same guarantee as the text path. Card commits tag source="matrix_card" for the audit log. - llm.chat_vision: multimodal /v1/chat/completions, same model, same gateway - spark.transcribe_card: faithful card->text, "" on a non-card (NONE sentinel) - bot.on_image/handle_card: download image, transcribe, hand to handle_intake - crm_client: source provenance overridable via the proposal's _source key - tests: test_spark.py + a provenance case; 41/41 suite green
Matrix intake bot
Turns a typed message in a dedicated Matrix room into a proposed fundraising-grid add/edit,
gated on in-thread human approval before any write. Runs as its own process (on the Spark),
separate from the CRM. Full design + rules: docs/guides/matrix-intake.md.
Run
# 1. Install the one third-party dep (isolated to this component — NOT the CRM runtime)
python3 -m pip install -r requirements.txt # matrix-nio
# 2. Fill the MATRIX_* and CRM_BOT_* vars in the repo .env (see ../../.env.example),
# and create a dedicated CRM user for CRM_BOT_USERNAME/PASSWORD (admin → invite user).
# 3. Start the listener
python3 bot.py
It primes the Matrix sync past history (no backlog replay), then listens. Post a message in the intake room; it replies in a thread with the parsed proposal. Reply yes to commit, edit field=value to change a field, or no to discard.
Layout
bot.py— entrypoint: connect, prime-then-listen, dispatch (lifts matrix-bridge's plumbing).parse.py— message → structured proposal via local Qwen (spark.py→backend/ingest/llm.py).proposals.py— in-memory pending-proposal store + the yes/edit/no state machine.crm_client.py— login +GET /api/intake/match+ write viaPOST /api/fundraising/log-communication.matrix_io.py— message splitting, thread-root detection, threaded-reply sender.settings.py— Matrix + CRM-API config (namedsettings, notconfig, to avoid shadowingingest/config).
Test (offline)
python3 test_parse.py && python3 test_proposals.py && python3 test_crm_client.py
# endpoint + create→match contract (boots the real server against a temp DB):
cd ../ && python3 test_intake_endpoints.py
Live Matrix behavior needs creds + matrix-nio and can only be smoke-tested on the Spark.