Index of every page on this site: llms.txt
All skills
deanpeters avatar

/pm-skills-navigator

@476ae12
Run it yourself

Find the right PM skill for the situation on your desk. Use when you don't know where you are in the product lifecycle or which skill in this library to start with.

Use this Skill: https://skilld.dev/gh/deanpeters/product-manager-skills/pm-skills-navigator

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

SKILL.md

≈46 tokens for metadata: the name and description. ≈5.6k when used: this file.

Description uses 2.3% of example budget

Before choosing a Skill, your Agent reads its name and description. All available Skills share that space.

  • A shorter description leaves more room for other Skills. This entry exceeds our 1% size suggestion.

In our Claude Code example, all Skill names and descriptions share 8,000 characters. This Skill uses ≈186 characters, or 2.3%.

The 1% threshold is a size suggestion. Longer descriptions can still fit.

Your model, settings, and other Skills decide how much text your Agent can read.

Example settings and source

The example uses a 200k-token context and default Claude Code settings. The count includes the name, description, separators, and when_to_use when present. Codex also counts local file paths.

Skit's source and limits: Codex 0.160.1, Claude Code 2.1.292.

PM Skills Navigator

Purpose

Get from "here's what's on my desk" to "here's the skill to open first, and why", in about five minutes. The library has dozens of skills. Most PMs don't need a catalog; they need a place to start and a sense of what comes next.

The navigator works by finding where the work is in the product lifecycle, because the most common mistake in product work is doing a later job before an earlier one is finished: writing the PRD for a problem nobody has validated, or prioritizing solutions before the outcome is named. The phase tells you which skills are premature.

It recommends a route, not a mandate. Every skill it points to runs on its own.

Input

Works best with: The situation on your desk, in a sentence or two: what's happening, and what you're supposed to produce or decide.

Also useful: What you already know for sure versus what you're assuming, who's asking, and how much time you have.

Anything supplied with the invocation itself (text after the skill name, a pasted context dump, or an appended ARGUMENTS: line) counts as answers already given. Use it and skip whatever it covers; don't re-ask.

Arriving empty-handed? That works too. The first question asks what's on your desk, and the rest are answerable from what you already know. "I don't know" is a valid answer and becomes a labeled assumption in the route.

Example invocations:

  • Our dispatchers don't use the route optimizer and my VP wants a redesign. Where do I start?
  • I just joined a product team and don't know which of these skills matters for me.

Key Concepts

The Arc

Product work tends to unfold in an order. The library groups its skills by that order:

Phase (bundle) The question You're here when... You can leave when...
Determine What outcome are we after, and what is the market doing? You can't state the outcome in a measurable sentence, or the market has moved You have a measurable outcome and a view of the market and the customer
Discover What is the real problem? You have an outcome but the problem is a guess or a stakeholder's opinion You have a problem statement backed by evidence from customers
Define Is this solution worth building, and how do we position it? You have a problem but the solution is untested You have a tested bet worth building and can say who it's for and why you win
Develop How do we turn it into buildable work? The solution is chosen but isn't ready for a team (no PRD, stories, order) A team can start, and knows what they're building and in what order
Deliver How do we price it, take it to market, measure it, and retire it well? It's built, or it's aging It's priced and launched, and you can tell whether it worked, or it has been retired cleanly
Drive How do I work with people and grow? The question is about a stakeholder, an incoming request, or your own career It's handled. This phase runs under all the others

The same arc shows up in software delivery (plan, analyze, design, build, deploy) and in other product frameworks. That's the point: it isn't a house opinion, it's how work earns the right to move forward.

The Phase Is Set by Evidence, Not by Where You Wish You Were

A team that says "we're in Develop" because a PRD is due, with an unvalidated problem underneath, is actually in Discover. Ask what's validated and what's assumed, and place the work at the earliest phase whose exit condition isn't met. Calling it later doesn't make it later.

A Cycle, Not a Line

Work loops back (a failed test sends you to Discover again) and runs concurrently (launch prep starts while you build). So the navigator:

  • Names two phases when the work straddles them, and puts the one carrying the highest unresolved risk first.
  • Treats Drive as orthogonal. A question about a stakeholder or your own career gets routed to Drive whatever phase the product is in.
  • Never blocks. If you want a later-phase skill, it says what you're skipping and what that risk is, then helps you anyway. Your call.

Motions: The Smaller Groups Inside a Bundle

Each bundle has motions, which are specific jobs. The map below is the full list. A motion marked Coming has no skill yet. The navigator says so instead of forcing a poor fit.

Anti-Patterns (what this is NOT)

  • Not a pipeline. It suggests an order; it doesn't require one. Skip, reorder, or stop after one.
  • Not a gate. It never refuses to help with a later-phase skill.
  • Not a replacement for the skills. It points; the skills do the work.
  • Not a catalog. For a flat list, see catalog/skills-by-bundle.md.

Facilitation Source of Truth

Use workshop-facilitation as the interaction protocol. Give the heads-up at the start (four questions, about five to ten minutes) and let the user dump context to skip ahead.

Default entry mode: Guided · set in pm-skills.config.yaml (library-wide default or a per-skill override); the user's own choice in the session always wins. On the first turn, offer all three modes, then open in the default without waiting for a pick: Guided asks the first question, Context dump invites the paste, Best guess asks for a one-sentence minimum and drafts with every assumption labeled.


The Map

This block is generated from pm-skills.config.yaml. Don't edit it here; run python3 scripts/check-bundles.py --write-navigator.

<!-- BUNDLE-MAP:START -->

Determine (Context): Set outcomes and read the terrain: strategy, finance, market, customers.

Discover (Discover + Investigate): Find and frame the real problem.

Define (Discover + Define): Shape a solution worth building and test the bet.

Develop (Create): Turn the chosen solution into buildable work.

Deliver (Deliver): Take it to market, measure it, and retire it well.

Drive (Runs under all five phases): Drive your career and your stakeholders.

Core (shared helpers that ship with every bundle): workshop-facilitation, pm-skills-navigator, autonomous-investigation, pm-skill-creator, skill-authoring-workflow

<!-- BUNDLE-MAP:END -->

Application

This skill asks up to 4 adaptive questions, then recommends a route with reasons.


Question 1 of 4: What's on your desk?

Agent asks:

"What's the situation? Tell me what's happening and what you're expected to produce or decide. One or two sentences is plenty."

Listen for the job-to-be-done behind the request. "Write a PRD" is a deliverable; the situation underneath (a stakeholder asked for a feature, or discovery found a pattern) decides which phase it belongs to. If the answer is about people (a stakeholder, a request, your career), note it: that's Drive, and it can skip Question 2.


Question 2 of 4: What's validated, and what's assumed?

Agent asks:

"Let's find where this really stands. Which of these is true today?

  1. We have a measurable outcome we're aiming at (not just a feature)
  2. We've talked to customers and can state the problem with evidence
  3. We've tested a solution and it held up
  4. None of those yet, or I'm not sure

Pick any that apply."

Place the work at the earliest phase whose exit condition is unmet:

They said Earliest unmet Phase
4 (none) The outcome Determine
1 only The problem Discover
1 and 2, not 3 The solution test Define
1, 2, and 3 Buildable work, then market Develop, then Deliver

If their situation sounds later than their answers (a PRD is due but they picked 4), say so gently: the PRD is premature, here's the evidence gap, and here's the smaller job that unblocks it.


Question 3 of 4: What do you need to walk away with?

Agent asks:

"What would make this session worth it?

  1. A decision (which option, build or not, which framework)
  2. An understanding (what's going on, what customers need, what the market is doing)
  3. A document (a PRD, stories, a positioning statement, a plan)
  4. Alignment (getting people on board, handling a request, planning a conversation)"

Use the answer to pick the motion and the skill type. A decision usually means an interactive advisor. An understanding usually means a workflow or an investigation. A document means a component skill. Alignment means a Drive skill, whatever the phase.


Question 4 of 4: How much time do you have?

Agent asks:

"Roughly how long can you give this: under an hour, a few hours, or days?"

Use it to size the route, and honor an override either way. Under an hour means one skill, the quick one. A few hours means two or three in order. Days means the workflow version where one exists. Name what a shorter route leaves uncovered.


Final Step: Recommend the route

Deliver this structure:

## Your Route: [Situation in a few words]

**Where the work is:** [Phase] (and [second phase] if it straddles) — because [what's validated vs assumed]
**Bundle / motion:** [Bundle] → [Motion]

### Start here
1. [`skill`](../skill/SKILL.md): [why this one first, one line]
2. [`skill`](../skill/SKILL.md): [what it adds]
3. [`skill`](../skill/SKILL.md): [optional; what it adds]

**What you're skipping, if anything:** [the earlier-phase job you aren't doing, and the risk]
**Gap, if any:** [a need with no skill yet, and the closest alternative]

### Paste this first
> [A real first prompt using their own details]

### Assumptions I made
- [Anything answered "I don't know"]

Then offer numbered next steps:

"Where next?

  1. Run the first skill now, with the prompt above
  2. See the rest of this phase, or the phase before or after it
  3. Check what you're skipping, and go back for it first
  4. Navigate a different problem

Pick a number, combine them, or tell me what you'd rather do."


Examples

Example A: SaaS, the PRD is premature

Fieldlight (field service software). The PM's VP wants a redesign of the route optimizer because dispatchers "don't use it."

Q1. "Dispatchers don't use the route optimizer. My VP wants it redesigned. I'm supposed to write the PRD."

Q2. Which is true: outcome, problem, tested solution? → "Honestly, none. The VP has an outcome in mind, but nobody has asked a dispatcher why."

Q3. What do you walk away with? → "A document, the PRD."

Q4. Time? → "A couple of hours."

Route: Phase Discover (and Develop, where the request is pointing). The PRD is premature: the problem is an opinion, not a finding.

  1. problem-framing-canvas: state the problem before the solution
  2. discovery-interview-prep: plan five dispatcher conversations
  3. prd-development: once the problem is evidenced

What you're skipping: nothing yet; you're not skipping, you're sequencing. Paste this first: "Frame this problem: dispatchers 'don't use' the route optimizer, and my VP wants it redesigned."

The coaching beat: the PRD was the right deliverable and the wrong next step. Naming it that way keeps the PM from feeling blocked.

Example B: Industrial, the work straddles two phases

Northfield Automation (industrial retrofit controls). NFA-200 sales have been flat for six quarters. A distributor asks for a variant; engineering says it's a 14-month cycle.

Q1. "NFA-200 is flat. A big distributor wants a variant. Sales is half-promising it."

Q2. → "We have an outcome (hold margin). We don't know if it's the product or the channel."

Q3. → "A decision: variant or not?"

Q4. → "A week, but it's a big call."

Route: Phase Deliver (an aging product) and Define (the variant is an untested bet). Highest unresolved risk first: Deliver, because nobody has established whether the product is declining or badly distributed.

  1. lifecycle-play-advisor: is this extend, replace, retire, or nothing yet?
  2. incoming-request-advisor: decode what the distributor is asking before replying
  3. pol-probe-advisor: if "extend" wins, test the variant cheaply

What you're skipping: a 14-month build commitment on an untested variant. Paste this first: "Help me decide what to do with NFA-200: sales flat six quarters, a distributor wants a variant."

The coaching beat: two phases named on purpose. Straddling is normal; what matters is which risk to retire first.


Common Pitfalls

Pitfall 1: Placing the Work Where the Deliverable Is

Symptom: "A PRD is due, so we're in Develop." Consequence: A polished spec for an unvalidated problem. Looks like progress, builds the wrong thing well. Fix: Place by evidence: the earliest phase whose exit condition is unmet.

Pitfall 2: Treating the Route as a Pipeline

Symptom: Running all three recommended skills in order because they were listed. Consequence: Process theater. The route is a suggestion; the first skill may be all you need. Fix: Stop when the question is answered. Every skill runs alone.

Pitfall 3: Forcing a Fit on a Gap

Symptom: Recommending a loosely related skill for a need the library doesn't cover yet. Consequence: The PM trusts a template that doesn't fit, and the library loses credibility. Fix: Say "no skill for this yet" and name the closest honest alternative with its limits.

Pitfall 4: Blocking the Skip

Symptom: Refusing to help with a later-phase skill until an earlier one is done. Consequence: Contradicts a cycle that really is concurrent, and the PM stops asking. Fix: Name what's being skipped and the risk, then help anyway.

Pitfall 5: Routing a People Problem Through the Product Arc

Symptom: Sending a stakeholder conflict to Discover. Consequence: Customer research for a problem that's about a person. Fix: Drive runs under every phase. Route people questions there directly.


References

Related Skills

These stand on their own; none is a prerequisite for this skill, and this skill isn't a prerequisite for them.

External Frameworks

  • Software development life cycle (plan, analyze, design, build, deploy): the same arc in engineering
  • Productside Blueprint: the product lifecycle the phase names are aligned with

Provenance

  • Distilled from practitioner experience teaching the product lifecycle and routing PMs to the right tool for the phase they are actually in.

Source: SKILL.md on GitHub

No rule matched.

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

No alerts7d3 checks · Risk SAFE
  • Gen Agent Trust Hub7d

    The PM Skills Navigator is a safe diagnostic tool that helps Product Managers identify the correct specialized skills for their current task. It uses an interactive questioning process to map the user's situation to a product lifecycle phase. The skill does not perform any risky operations such as network requests, code execution, or sensitive file access.

  • Socket7d

    No alerts

  • Snyk7d

    Risk: LOW · No issues

Signed by skilld at 476ae12. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub last week.

Activeupdated last week
argument-hint
[the situation on your desk, in a sentence or two]
type
interactive
theme
navigation
Other metadata
intent
Guided triage from a situation to a route. Places the work on the product lifecycle by what is validated versus assumed, names the bundle and motion that fits, recommends one to three skills in order with a reason each, says plainly when the library has no skill for the need yet, and hands over a paste-ready first prompt.
best_for
[
  "Not knowing which of the library's skills fits the problem on your desk this week",
  "Checking whether you are about to do a later-phase job (a PRD, a roadmap) before an earlier one is done",
  "Learning how the library's skills fit together by lifecycle phase"
]
scenarios
[
  "Leadership wants a feature and I can't tell if we should be framing, testing, or specifying it",
  "I'm new to this library and want to know where to start for my product"
]
estimated_time
5-10 min

README badge

README badge for deanpeters/product-manager-skills/pm-skills-navigator