Sniff Test — Portfolio Add-On

Outcome

Across a set of initiatives (a portfolio Kanban or any backlog), surface the board-level patterns a single-initiative sniff can't see — so leadership can rebalance risk, kill duplication, and find hidden WIP. Per-initiative clarity reads come from the base sniff-test skill; this add-on is about the relationships between items.

When To Use

Reviewing a whole portfolio Kanban or backlog, not one item.
Prepping a sponsor / ELT portfolio review or a "where are we over-invested?" conversation.
After running single-initiative sniffs on several items and wanting the cross-cutting story.
Checking whether the portfolio's risk exposure is intentional.

Prerequisite

Run (or assume) base sniff-test reads on the individual items first — verified stage + clarity-by-dimension per initiative. This add-on consumes those reads; it does not replace them.

How to Run Me — What to Feed Me

Point me at the portfolio's system of record: the Portfolio Kanban (Tier-1 board), the Jira project/board, the Confluence portfolio space, or the folder where initiative canvases and roadmaps live. The more of each initiative's Lean Product Canvas, Outcome-Oriented Roadmap, and recent sponsor update I can reach, the more reliable the stale-status sweep and risk-balance split become. With thin input I'll work from whatever board snapshot you give me and flag low coverage — board-level findings degrade fast when the per-initiative evidence isn't reachable.

Scan Approaches — pick deliberately

Approach	What it is	When to use
Full-board baseline	Sniff every initiative as a baseline	A dedicated session or major portfolio-review agenda item. Establishes the baseline you compare actuals against later.
Random sample	Sniff a random selection across stages	Building the muscle; spot-checking board health.
Stage debrief	Sniff all items that just passed a stage	e.g., everything newly committed into Execute/Build.
Cadence	Sniff on a rhythm	After a planning interval; as part of the Strategic Portfolio Review.

State the chosen approach and the altitude (Tier-1 enterprise only, or including Tier-2 domain/team work) in one line.

Portfolio Patterns To Detect

Duplicate / shadowing initiatives. Two+ items covering substantially the same problem, or one quietly subsuming another. Smell: fuzzy boundaries between a parent initiative and its "refinement." Move: merge, make one a sub-issue, or sharpen the distinct bet each represents.
WIP clustering / hidden imbalance. Many initiatives piled in one domain or one stage while others starve. Smell: every "top priority" pointing the same way; lots of work stuck just before a commitment point. Move: surface the real tradeoff for a leadership capacity decision.
Portfolio risk balance (Explore/Discovery vs. Build & Deliver). What share of capacity sits in Explore / Discovery (uncertain, exploratory, smaller bets) vs. Execute/Build + Rollout (committed, known ROI trajectory)? Frame this like a financial portfolio — leadership should see the split and manage it intentionally (one enterprise engagement used ~20% in de-risking / ~80% in build-and-deliver as a starting reference point — treat it as an anecdote to calibrate against, not a benchmark). Smell: ~100% committed work (no exploration → future surprises) or runaway exploration that never commits.
Stale-status sweep. Board-wide version of the single-item check: which status labels disagree with the evidence (shipped output, downstream artifacts, go-live state, git history)? Stale labels distort the whole portfolio picture. Move: correct the stages, then re-read the balance.
Experiment scope creep. Items framed as small Explore/Discovery experiments that have ballooned into full builds. Smell: an "experiment" with a delivery-sized footprint and no proceed/stop decision. Move: re-cut to the riskiest assumption, or graduate it honestly to a committed Execute/Build.
Watermelon clustering. A pattern of initiatives that were green throughout and turned out misplaced — points at a systemic gap in how the portfolio reads its own confidence, not just one bad bet.

Rules

Patterns, not re-grading. Don't re-sniff each item in depth; consume the base reads and report the relationships.
Correct stages before reading balance. A stale-status sweep comes before any Explore/Discovery-vs-Build&Deliver split, or the split is fiction.
Make tradeoffs visible; don't resolve them silently. The output of WIP/balance findings is a leadership decision surfaced, not a reshuffle done unilaterally.
Speak portfolio risk in financial-portfolio terms and in maturity/confidence language, not "phase gates."
Keep altitude honest. Only Tier-1 enterprise bets belong on the portfolio Kanban; flag Tier-2 domain/team work that's been escalated without meeting the criteria, and vice versa. The criteria are in sniff-test under "Tier-1 or Tier-2" — use the organization's own if it has them, and name which you applied. Do not assess against criteria you invented and did not state.

Output Modes

portfolio scan (default): a compact table of initiatives × verified stage × clarity summary × pattern flags, plus a short narrative of the 2–4 cross-cutting findings and the risk-balance split.
risk-balance view: just the Explore/Discovery vs. Build & Deliver split with the rebalancing conversation.
review brief: a sponsor/ELT-ready summary — what's duplicated, where WIP is hidden, whether risk exposure is intentional, and the decisions requested.

Quality Gates

A stale-status sweep is done before any risk-balance claim.
At least the duplication, WIP-clustering, and risk-balance patterns are explicitly checked.
Each pattern flag names a specific leadership conversation or decision, not a vague observation.
The risk-balance split is stated, with whether it looks intentional.
Altitude (what belongs on the portfolio board) is assessed.

References

Base skill: sniff-test (../sniff-test/SKILL.md) and its references/ for per-initiative clarity dimensions, the lifecycle rubric, and the Risk/Derisking Matrix.

Source

If someone asks why a rule here exists and you can browse, fetch Scaling Product Organizations with Portfolio Agility and answer from it rather than paraphrasing — the reasoning is there and it is better than your summary of it. Never required: this skill runs fully offline.

The multi-initiative add-on to sniff-test. Same lineage: distilled from Scaling Product Organizations with Portfolio Agility by Yuval Yeret, building on Mark Richards' portfolio-agility work and the Lean Product Canvas by Jeff Gothelf and Josh Seiden. Full credits and links to the originals: CREDITS.md.

These are Yuval's questions, not his judgment — don't present the output as his read of your situation.