fb-discovery

name	fb-discovery
description	Use when product unknowns need research, experiments, competitor evidence, opportunity evidence, or feasibility evidence before implementation is approved.

FB Discovery

Discovery reduces uncertainty for Product/User. It is a planning/evidence workstream, not implementation. Research the smallest decision-changing unknown, record observable evidence, and hand a recommendation to Product/BFM. Discovery must not implement source, present speculation as evidence, or set final Product priority.

Use records.md: research decisions and limits stay in the task handoff, raw bounded proof stays in the QA artifact, and the Discovery card carries only current task IDs, blockers, next action, and links.

For a known task and concrete question, call MCP fb_project_context first and open only its relevant cited sources. The graph routes to authoritative records; it is not a source of truth. Use the board → index → handoff → card fallback when the packet says fallback or is incomplete or contradictory.

Mini-loop

1. Read AGENTS.md, PROJECT_BOARD.md, docs/handoffs/index.md, the linked handoff, docs/workstreams/fb-discovery.md, and relevant existing evidence.
2. State the unknown, decision it blocks, hypothesis, success/failure signal, scope, and timebox before research or an experiment.
3. Gather the smallest useful research, experiment, competitor, opportunity, or feasibility evidence. Label direct findings, inference, and unresolved hypotheses separately.
4. Compare evidence with the approved Product goal. Recommend proceed, change approach, run another bounded experiment, defer, or stop.
5. Create or update docs/handoffs/<TASK-ID>.md; do not edit application source, branch, commit, submit, merge, deploy, or change provider state from ordinary Discovery chat.

Ready handoff contract

Use frontmatter Product/BFM can scan:

---
type: fb-lane-handoff
task: <TASK-ID>
lane: fb-discovery
status: ready
---

Use status: blocked when required evidence cannot be gathered. Do not mark a hypothesis or an unrun experiment ready as if it were a finding. The handoff must include the unknown and decision, method and sample/source limits, findings, hypotheses/inferences, competitor and opportunity evidence when relevant, feasibility evidence when relevant, risks, recommendation, and next owner.

For non-trivial work include:

## Goal Alignment Session

Product Goal: <approved Product/workstream goal>
Workstream Goal: <decision-changing unknown>
Lane OKR Fit: aligned | suggest approach change | blocked by OKR ambiguity
User Approval Needed: yes | no
Mini-loop Evidence: <observable research or experiment evidence>
Evidence Against Product OKR: <counter-evidence> | None identified

Before closeout, re-read the unknown, evidence, recommendation, and Product goal. If evidence is insufficient, report the exact gap instead of manufacturing confidence. Product/BFM owns prioritization, integration, and source execution.