When to use
Use this to open and run the file for one matter — a specific launch, feature, or campaign under review — so every intake fact, finding, condition, and counsel decision lives in one place. It is the connective tissue: the other product-legal skills read from and write to this workspace, and it becomes the audit trail if a decision is later questioned.
Not for: company-wide configuration (use prodl-customize), producing the actual verdict (use prodl-launch-review), or spotting issues (use prodl-is-this-a-problem).
Method
- Open the matter. Assign a name, owner, go-live date, and the
prodl-customizeprofile it runs under. - Seat the intake. Drop in the
prodl-cold-start-interviewrecord as the fact base. - Register findings. Each finding from a feature or claims review lands here with its tier, owner, and status.
- Track conditions to closure. Every go-with-conditions item is open until its owner marks it done with evidence.
- Log the escalation trail. Record what went to counsel, when, and the disposition — the paper trail.
- Hold the verdict and its basis. The final go/no-go plus the findings that drove it.
- Set the review trigger and archive. On go-live, snapshot the state; on the trigger event, reopen.
Example
Matter "Q3 Sleep-App Launch" opens with the intake fact sheet, then accumulates: a claims finding (unsubstantiated "2x faster", Unlawful, owner Marketing, open), a data finding (undisclosed SDK, High, owner Legal Ops, closed with updated policy), and a counsel note approving revised copy. Verdict recorded: Go-with-conditions. Two weeks post-launch a new claim is added — the matter reopens per its review trigger instead of starting cold.
Pitfalls
- Scattering the record. Findings in Slack, conditions in a doc, counsel notes in email means no trail — keep one workspace.
- Closing conditions without evidence. "Done" needs the artifact (updated copy, study, disclosure), not a checkbox.
- No escalation log. If you can't show what counsel saw, you can't show the decision was informed.
- Never archiving. An open matter with a shipped launch rots; snapshot at go-live.
Output format
Matter: <name> — owner — go-live — profile
Intake: <linked fact sheet>
Findings:
- <issue> — tier — owner — status (open/closed) — evidence
Conditions outstanding:
Escalation log: <date — item — counsel — disposition>
Verdict + basis:
Review trigger / archive state:
Reference
FTC substantiation. The workspace is where substantiation lives: for each objective claim, store the evidence (or its absence) on the record. "Reasonable basis before dissemination" is auditable only if the proof is filed against the claim it supports.
Launch-risk rubric. The matter's overall status = worst open finding. Any open Critical keeps the matter at No-go; the workspace should surface that at a glance rather than burying it in a list.
When to escalate to counsel. The workspace does not decide — it records. It routes every Critical, Severity-5, and regulated-category finding to the escalation log for an attorney, and holds counsel's disposition. This skill is organization and audit trail, not legal advice; the escalation trail exists precisely because an attorney, not this workspace, owns the calls on flagged items.