Bubble-invest

fund-catalyst-calendar

Build and maintain a forward-looking catalyst calendar for holdings + watchlist names. Covers earnings dates, corporate events, regulatory decisions, macro releases. Use during morning catalyst-prep, pre-earnings positioning, or when the principal asks "what's coming up next week/month." Triggers on "catalyst calendar", "upcoming events", "earnings dates", "what's coming up".

fund-research-posture

Research methodology + analytical-tooling standard for the agent. Load before composing a thesis, proposing a position, or building new analytical code. Anchors backtest honesty, library use, and citation conventions.

fund-thesis-format

Canonical 7-part investment thesis structure. Use whenever the agent writes a position note, proposal memo, decisions.reasoning row, theme note, or a brief naming a specific bet. Every new thesis MUST follow this skeleton.

fund-thesis-scorecard

Maintain a running scorecard tracking each thesis pillar's status over time. Pairs with fund-thesis-format. Use to systematize the "is my thesis still intact?" question — converts narrative review into a structured pillar-by-pillar trend read. Use on a weekly thesis revisit, when a major data point lands (earnings, regulatory, macro), or when the principal asks "is the [TICKER] thesis still working?".

DevOps & Infrastructure Listed

cloud-wiki-compile

VPS-side single always-on compiler for the shared wiki. Three modes — compile (nightly transcript mining), synthesis (weekly thesis), pruning (weekly TTL/cap maintenance). Mines 6 VPS agents + both Mac caches.

broker-adapter

Generic broker interface for the example portfolio agent. Documents the adapter pattern (get_positions / get_nav / submit_order) with an allow_live poka-yoke, a no-op stub for demos, and Alpaca shown as one worked example using env-var credentials. Plug your own broker SDK behind this interface — NO real keys ship here.

fund-source-verification

Source-tier classification + independent-verification rubric for any external signal (news, newsletters, paywalled wires, sentiment trackers). Load whenever you're about to act on a third-party claim or write "<outlet> reports X" in a memo.

department-onboarding-guide

Conversational onboarding skill for new bubble-ops-loop departments. Guides operator through the 7 eclosure steps (Mandate -> Missions -> Layers -> Skills/tools -> Gates/KPIs -> Dry-run -> Activation), produces schema-validated artifacts at each step, manages the onboarding branch + activation PR. Use when bringing up a new dept (Maya, Ben, Tony, Miranda, Eliot, etc.). The skill ships templates + state machine + dry-run simulator; the agent (Claude Code session) consumes it to drive the conversation.

dept-spawner

Spawn a new Bubble ops department (agent) end-to-end — create the repo, scaffold the dept tree, wire the VPS prerequisites, and hand off to the self-driving onboarding. Use this whenever the user wants to create/spawn/hatch/onboard a NEW agent or department (like Ben the fund manager or Geraldine the accountant), set up a new dept's repo + systemd unit + Telegram bot + secrets, or asks "add a new agent/dept", "spin up a department for X", "onboard a new ops agent". This orchestrates the EXISTING scaffolding (bootstrap-dept.sh + scaffold.py + the self-driving onboarding skill) — it does not reinvent the process. Do NOT use for editing an existing live dept (that's a normal PR) or for non-agent infra.

emit-kanban-task

Push an action-required item onto the Bubble ops board (the GitHub-Issues control plane Bubble-invest/bubble-ops-board, rendered in the cockpit /kanban) so a human or agent can triage it later — instead of letting a finding evaporate in chat. Use whenever you uncover something that needs follow-up but is OUT of the current scope: "add this to the kanban / backlog / for tomorrow", "flag this for {{OPERATOR}}", "we should track X", "create a card for…", "remember to do Y later", or any decision/approval/incident/finding that shouldn't block the current task but must not be lost. Reach for this the moment you think "that's worth doing later" — that's the signal.

pre-publish-scan

Systematic pre-open-source LEAK GATE for any Bubble repo. Run BEFORE making (or while a repo is) public, and at each step of an open-sourcing effort, to catch operator PII ({{OPERATOR}}/{{OPERATOR_2}} Telegram IDs), internal infra (Tailscale FQDN/IP, Mac paths), internal Notion IDs, secret-map docs, and raw secret prefixes — in the working tree AND git history. Use whenever you're about to publish, push to a public repo, open-source a skill/plugin/repo, run a launch, or whenever {{OPERATOR}} says "is this clean to share / can this go public / scan for leaks". This is the GATE: a repo does not go public until this exits clean.