| name | orientation |
| description | Audit product identity — what we do, for whom, whether the feature surface is coherent with the thesis, and whether context files agree with each other. Use when the product feels unfocused, after a pivot, before strategic planning, or when something in the product story feels off. |
| user-invocable | true |
| argument-hint | [area or question] |
MANDATORY PREPARATION
Invoke /product-thinking — it contains the Context Gathering Protocol and the AI Slop Test. Follow the protocol before proceeding.
Mindset
Most products drift. Features accumulate because customers asked, competitors shipped, or someone had a slow week. Over time, the product says it does one thing but actually does twelve — and none of them well enough to be the reason someone chooses it.
Orientation is the audit that catches drift. You are not here to validate. You are here to hold a mirror up and ask: does the product we built match the product we say we are?
Context Pull
Load everything. This is a read-heavy skill:
- Product context. Read
.acumen.md — positioning, thesis, stage, target users, strategy.
- Feature inventory. Read
.acumen/features.md — every capability the product has today.
- Personas. Read
.acumen/personas.md — who we say we serve, and their behaviors.
- Value chain. Read
.acumen/value-chain.md — the end-to-end workflow per persona. Use this to check whether features align with the steps where we claim to bring value.
- Competitors. Read
.acumen/competitors.md — how the market frames us vs. alternatives.
If any context file is missing, flag it. Orientation without a feature inventory or personas is guesswork.
Core Protocol
1. Thesis Check
State the product thesis in one sentence. Then ask:
- Can every feature in the inventory be traced back to this thesis?
- Which features serve the thesis directly? Which serve it indirectly? Which don't serve it at all?
- If you removed the orphan features, would anyone notice? Would retention change?
Features that don't connect to the thesis are either: (a) the thesis is wrong and needs updating, or (b) the feature is drift and should be deprecated or spun out.
2. Persona Alignment
For each primary persona in .acumen/personas.md:
- Which features serve this persona's core job?
- Which features are irrelevant to them (clutter)?
- Is there a gap — a job this persona needs done that the product doesn't address?
- Value chain coherence — Do the features we built match the steps we claim to own in
.acumen/value-chain.md? Are there chain steps marked as "Own" with no corresponding feature, or features with no chain step?
- Would this persona describe the product the same way the positioning does?
If your primary persona would describe the product differently than your marketing does, you have an orientation problem.
3. Surface Coherence
Look at the feature surface as a whole:
- Breadth vs. depth — Is the product wide and shallow (many features, none best-in-class) or narrow and deep (few features, each excellent)? Which should it be given the stage and strategy?
- Entry point clarity — If a new user lands in the product, do they understand what to do first? Does the product guide them toward the core value or present a wall of options?
- Feature relationships — Do features reinforce each other (composable) or exist independently (bolted on)? Composable features compound. Bolted-on features create maintenance burden without compounding value.
4. Positioning Test
Take the positioning statement from .acumen.md. Test it:
- Exclusion test — Does the positioning exclude competitors? If any direct competitor could use the same sentence, it's a category description, not positioning.
- Feature test — Could someone use the product for a week and confirm the positioning? Or is it aspirational?
- Persona test — Would the primary persona use these words to describe the product to a colleague?
5. Context Coherence
Context files are written by separate skills at different times, so they drift. Before trusting the audit above, cross-check that the files agree with each other. This is a fact-check, not a rewrite — surface disagreements, do not edit any file.
Run these checks; skip any whose inputs are missing:
- Persona references — every persona named in
personas.md should be referenced in features.md, value.md, and value-chain.md. Flag orphan personas (defined but never referenced), phantom personas (referenced but not defined), and name drift (same persona under slightly different names across files).
- Feature references — every feature in
features.md should serve a persona that exists in personas.md. Flag personaless features and featureless personas.
- Metric references — every metric named in
value.md ("how we know it's delivered") should have a source in sources.md that could produce it. Flag unsourced metrics and metric drift (same concept, different names).
- Value vs. features — features listed in
value.md as delivering value should actually appear in features.md. Flag value claims without features and unvalued shipped features.
- Strategy vs. features — areas called out in
.acumen.md strategy should have corresponding features; major feature areas should be reflected in strategy. Flag strategy gaps and silent feature areas.
- Competitor positioning — flag obvious disagreements between
.acumen.md positioning and "we beat them on X" claims in competitors.md.
Order findings by severity: contradictions first (real disagreement), then phantoms (broken refs), then orphans, then name drift, then coverage gaps. Cap each kind at 10 — if more, append "(+N more, run individual context skills to address)". Empty sections are fine and expected on a healthy project — do not manufacture findings to fill space.
Output Format
Product Thesis
One sentence. Pulled from .acumen.md or synthesized from context.
Orientation Score
| Dimension | Assessment | Evidence |
|---|
| Thesis coherence | Strong / Drifting / Misaligned | How many features connect vs. orphans |
| Persona alignment | Focused / Scattered / Missing personas | Which personas are well-served vs. underserved |
| Surface coherence | Composable / Fragmented / Cluttered | Whether features reinforce or dilute |
| Positioning accuracy | Accurate / Aspirational / Outdated | Whether the product matches its claim |
| Context coherence | Consistent / Minor drift / Contradictory | Findings from the Context Coherence check |
Context Coherence Findings
Group by kind. Omit any kind with zero findings.
- Contradictions —
[type] . Files: <file:section> ↔ <file:section>. Resolve with /<skill>.
- Phantoms —
<entity> "" referenced in <file> but not defined in <source file>. Action: define via /<skill> or remove the reference.
- Orphans —
<entity> "" defined in <file> but referenced nowhere. Action: link it or remove it.
- Name drift — "" in
<file> appears to be the same entity as "" in <file>. Pick one canonical name.
- Coverage gaps —
<Persona|Strategy area> has no <features|metrics|value mapping>. Run /<skill>.
If everything checks out, replace this block with a single line: "Context files are consistent across personas, features, value, sources, and strategy."
Feature Map
| Feature | Thesis connection | Primary persona served | Assessment |
|---|
| Direct / Indirect / Orphan | [persona name] or None | Keep / Evolve / Deprecate / Investigate |
Drift Zones
Where the product has drifted from its thesis. Be specific — name the features, the personas they serve (if any), and why they exist.
Gaps
Where the thesis promises something the product doesn't deliver. Missing capabilities that the positioning implies should exist.
Recommendations
Prioritized list of actions. For each:
- What: the specific action
- Why: what it fixes in the orientation
- Suggested next step: which Acumen command to run (
/workshop, /increment, /roadmap)
Do not pad. If the product is well-oriented, say so in two paragraphs. The value of this skill is honesty, not volume.