CWP

[Pipeline·OWNER] Decision #1 RESOLVED: taxonomy base confirmed as custom cwp_sector layer over public-domain SIC/NAICS ranges (GICS skipped, no licensing dependency). The taxonomy can be frozen once drafted. Remaining owner decisions: #2 crosswalk approval pass, #3 adjudication surface, #4 materiality floor, #5 pilot member list.

2026-08-06 · mcp

[Pipeline·OWNER] Five owner decisions for the conflict pipeline: (1) confirm SIC/NAICS taxonomy base [gates crosswalk drafting]; (2) crosswalk approval pass ~2-3hrs [gates the join, mid-Project-A]; (3) adjudication surface — recommended: spreadsheet in/out [gates pilot run]; (4) materiality floor value [gates publishing]; (5) pilot member list, 10-15, bipartisan-balanced [gates pilot run].

2026-08-06 · mcp

[Pipeline·C] API/Report Layer & Pilot Run: implement the 7 v1 endpoints (committees, assets, conflicts, wealth-summary, report-card, sectors/crosswalk, review-queue) with provenance fields (taxonomy/rule/resolver versions + source-PDF pointers); run pipeline over pilot members; adjudicate residue only. Gated on A+B complete. Acceptance: report-card endpoint renders a complete pilot card with zero hand-assembled data; zero membership errors.

2026-08-06 · mcp

[Pipeline·B] Resolution & Conflict Engine: Tier 1/2 asset→entity cascade (ticker → exact name → fuzzy → LLM → manual queue) with confidence + method + version on every row; spreadsheet in/out for the review queue; conflict join + materiality config (floor value = placeholder until owner decides). Can run parallel to A; join blocked on approved crosswalk. Acceptance: ≥90% non-private asset value auto-resolves; determinism check passes.

2026-08-06 · mcp

[Pipeline·A] Taxonomy & Crosswalk: freeze cwp_sector taxonomy over SIC/NAICS ranges; ingest committee/membership data from /srv/app-dev/data/cmtes into normalized schema (bioguide-keyed, subcommittee-level, ~2006-forward); LLM-draft ~250-row committee→sector crosswalk; OWNER approval pass gates completion. Validated against Congress.gov spot-checks.

2026-08-06 · mcp

[Launch] Define success metrics + analytics instrumentation before go-live: page traffic per district, newsroom pickups, social engagement, inbound corrections. Instrument first so you can tell whether district-first distribution actually reaches voters.

2026-08-04 · mcp

[Launch·Decision] Choose the launch shape: staged pilot (a handful of high-signal / event-driven districts, heavily QA'd) vs. all-435 big-bang. Your own "one debunked number poisons all 435" principle argues for a tight pilot. This decision reorders most of the other GTM work.

2026-08-04 · mcp

[Influencer] Build the influencer/partner target list + outreach approach. Pick categories that stay inside the nonpartisan constraint (accountability & data-journalism accounts, good-government orgs, local/civic creators) rather than partisan amplifiers.

2026-08-04 · mcp

[Social] Design the social pipeline, as automated as possible, on top of SocialPoster (project #12) rather than a new build. Resolve the tension in the brief: event-triggered (new filings) vs. the steady cadence social algorithms reward. Define the content unit (per-district card vs. national thread).

2026-08-04 · mcp

[Media·Decision] Newsroom pitch + press kit, and decide the embargo/exclusive model: do local outlets get an exclusive first crack at their own member's data? Exclusivity often earns the pickup, but complicates a simultaneous launch — resolve alongside launch shape.

2026-08-04 · mcp

[Media] Build the local-media target list (per-district outlets; start with pilot/launch districts, expand from there). Decide the sourcing method and data structure so it can scale toward all 435.

2026-08-04 · mcp

[Web] District-page SEO plan targeting real voter search intent (e.g. "[member name] net worth / finances", "[district] representative wealth"). Programmatic SEO is the low-cost organic distribution engine that makes district-first delivery work.

2026-08-04 · mcp

[Web] Auto-generated district landing-page system: templating spec + URL scheme + programmatic-SEO structure for up to 435 pages. Build ON TOP of WebBuilder (project #1) as the generator rather than a new stack. Handle member turnover / redistricting in the URL + data model.

2026-08-04 · mcp

[Web] Main site mockup — methodology, its published limits, corrections policy, about/funding, and the core pitch. Auditability-as-identity means the caveats/methodology have to be first-class, not footnotes.

2026-08-04 · mcp

[Product] Launch-scope data-completeness audit: enumerate which districts/members have publishable, fully-verified data vs. under-review or excluded years. Defines who you CAN launch, and feeds pilot selection.

2026-08-04 · mcp

[Product] Design the District Report Card artifact — the core creative unit everything else derives from (layout, page-level source-PDF links, disclosed-range/caveat display, "explained vs. unexplained growth" framing). Landing pages, social, and press kit all reuse this.

2026-08-04 · mcp

[Product·Decision] Define the minimum analytic set for a publishable District Report Card v1: is Wealth Growth Score enough, or does "excess growth over what the market explains" (the stated headline metric, currently planned-not-built) need to ship first? This answer drives the Report Card design.

2026-08-04 · mcp

[Legal·GATE] Election-law counsel sign-off before ANY paid or geo-targeted amplification. Hard gate — nothing paid ships until this clears. (Distinct from the defamation review.)

2026-08-04 · mcp

[Legal·GATE] Pre-launch defamation / publication-model review for claims about named members, plus a documented, operational corrections-and-takedown workflow (who responds, how fast, what the process is when a member's office disputes a number). Source-linking + ranges are the defense; get a lawyer's read on the model before publishing.

2026-08-04 · mcp

[Legal·Decision] Resolve legal entity / fiscal-sponsorship / 501(c)(3) status (open [OWNER] item in the funder docs). Determines what communications are permissible and how the PAC relates to the public brand.

2026-08-04 · mcp

[Identity] Stand up sending domain + email infrastructure (SPF/DKIM/DMARC, warmup, deliverability) before any newsroom or influencer outreach — cold pitches from a fresh domain land in spam.

2026-08-04 · mcp

[Identity·BLOCKER] Once the name is locked: secure the domain (primary + defensive variants + .org) and reserve social handles across all platforms in one pass (X, Bluesky, Instagram, TikTok, YouTube, Facebook, LinkedIn).

2026-08-04 · mcp

[Identity·BLOCKER] Decide the public-facing brand name and decouple it from "Blackeye PAC." Confirm "Congressional Wealth Project" (or choose an alternative) as the neutral, nonpartisan public brand; keep the PAC in the background so newsroom credibility and the nonpartisan-by-construction positioning aren't compromised. Gates domain, handles, email, press kit.

2026-08-04 · mcp