argen/hornero

Use when a team needs to decide which acquisition channel to invest in given the product's stage, market, and motion — paid / content / product-led / sales / community / partner — and wants a ranked recommendation with explicit tradeoffs rather than a generic list of options.

Use when a PM suspects their team is using AI tools without meaningfully changing how they work — to assess maturity across five AI-PM competencies and identify the highest-leverage capability gap to close first.

Use when diagnosing why a PM-to-Director transition is stalling — or when a newly promoted Director needs to map the two axes that define the role shift (scope and time horizon) and build clarity for their team from incomplete organizational direction.

Runs a rot-review pass that surfaces skills, roles, and hooks that may no longer be earning their keep. Run quarterly. Reuses validate-skill, smoke-check, and the eval suite to separate validated, still-load-bearing items from compensatory ones that are candidates for retirement.

Makes the implicit bets in a product strategy explicit: typically 3-7 bets, each with an assumption, evidence weight, kill criterion, and revisit cadence. Use after a strategy exists and the question is what must be true, and when to stop. To write or critique the strategy itself use strategy-kernel; to track evidence against one open hypothesis use hypothesis-track.

Produces a structured business-health read across four dimensions (growth, efficiency, retention, runway), integrating the SaaS metrics into one verdict with the weakest dimension named. Use before a strategy debate, board prep, or quarterly review when the question is how healthy the business is. It applies the metrics rather than teaching them: for explanations of growth and retention metrics use saas-revenue-growth-metrics, for efficiency metrics use saas-economics-efficiency-metrics, for a definitions lookup use finance-metrics-quickref.

Shows what Hornero ships (roles, cross-role skills, and per-role skills) as a scannable index. Use when the user wants to know what is available without reading the full files.

Queries the project's chronicle/entries/ layer for prior decisions, hypotheses, stakeholders, and customers whose scope, role, status, or keywords overlap the current work, and surfaces them before fresh analysis begins. Use before reviewing, proposing, or producing a non-trivial artifact in a project with a chronicle. If the project only has memory/decisions/ files use memory-query; to write a new entry use chronicle-record.

Captures a decision as a schema-validated entry in the project's chronicle/ layer (proposed via chronicle.sh, committed by a human) so future sessions and roles can load it as situational awareness. Use after a non-trivial decision in a project that has a chronicle/. It supersedes decision-record, which is the lightweight memory/decisions/ variant for projects without a chronicle; to read entries back use chronicle-recall.

Use when a PM needs to produce a structured research pass on a company — competitor, acquirer, partner, or prospect — that goes beyond surface-level web search and yields a defensible brief with source discipline.

Use when a PM needs to convert "we should look at competitor X" into a defensible position on what to build, refuse to build, and watch — structured analysis of a competitor's product, positioning, packaging, switching costs, and observable strategy.

Use when an AI workflow is producing vague, inconsistent, or expensive output — to diagnose whether the root cause is context stuffing and to redesign the information architecture grounding the AI.

Use when a session feels stuck — detects context drift, repeated rework, contradictory outputs, and recommends resetting with a fresh tight brief.

Runs cross-functional work as the PM orchestrator: frames a launch, launch-readiness review, contested decision, roadmap call, or strategic question as a customer job, dispatches the right role mix, and synthesises the findings into a product decision with outcomes, success criteria, the alternative considered, and a revisit condition. Use when asked to critique, pressure-test, or sign off a launch plan, or when a question crosses product, engineering, design, marketing, or data. For a single feature spec use prd; for the strategy itself use strategy-kernel.

Use when a team needs to lay out the end-to-end customer journey across stages, touchpoints, emotions, and pain points to find the highest-impact intervention.

Use after a meaningful event on a B2B customer account — new metric snapshot, support escalation, SLA event, expansion signal, or a new contact discovered. Appends to the relevant section on the type: customer chronicle entry (Usage shape, Health signals, Decision drivers, Stakeholders) via chronicle.sh append, keeping the account picture current without forcing a full supersede.

Captures a decision as a lightweight document in memory/decisions/: the diagnosis, the decision, alternatives rejected and why, the validation plan, and the revisit-by date. Use after a non-trivial decision in a project without a chronicle/ layer. If the project has chronicle/ use chronicle-record, the strict-schema successor; to read prior decisions back use memory-query.

Use when a PM is preparing for, actively interviewing for, newly settled into, or struggling to recalibrate within a Director of Product role — to surface the specific failure modes and highest-leverage moves for that exact situation.

Produces a discovery interview guide before the call, grounded in Teresa Torres' continuous-discovery methodology and Bob Moesta's switch-interview technique. Use when asked to plan, script, or prepare for a customer interview. After the interviews, turn transcripts into snapshots, themes, and a quote bank with interview-synthesis.

Use when entering a new VP or CPO role — to structure the first 90 days as a diagnostic process that builds the evidence base before making structural, strategic, or people decisions.

Runs automatically at session end via the Stop hook (or invoke explicitly mid-session after a notable success/failure) — scans the conversation for surprises and corrections, proposes durable updates to roles, skills, CLAUDE.md / AGENTS.md, or per-project memory.

Use when a PM needs to evaluate a feature-investment decision through a finance lens — fully-loaded cost, opportunity cost, expected value, payback — not vibes or ICE-without-finance.

Use when a PM needs to set or change pricing using finance levers — cost-plus, value-based, willingness-to-pay, margin floors, packaging — not just "what competitors charge"

Produces a one-page lookup of the ~30 SaaS finance metrics a PM is expected to read, each with definition, formula, healthy range, and common misread. Use when asked for a cheat sheet, a quick definition, or prep for a board or CFO conversation. For depth on growth and retention use saas-revenue-growth-metrics; for depth on unit economics and efficiency use saas-economics-efficiency-metrics; for a diagnosis of the actual business use business-health-diagnostic.

Use to audit whether a repository is AI-ready — does it have AGENTS.md / CLAUDE.md, well-scoped skills, useful test feedback, and a working install path.

Use when a PM needs a current read on the GTM pipeline — stage-by-stage health, deal velocity by segment, ICP-fit pattern of who's actually moving — without waiting for a Tuesday sales-ops sync. Reads HubSpot via MCP; produces a one-page readout that names what's moving, what's stuck, and the one question the data raises.

Use when new evidence arrives on an open type: hypothesis entry — appends a dated row to its Evidence Log via chronicle.sh append, then either keeps the hypothesis open or surfaces a threshold crossing (falsifier triggered → propose refutation; prediction met → propose a resolving decision via supersedes).

Use to feed an artifact (interview transcript, meeting notes, doc export, dashboard screenshot, support ticket, paste) into the project's chronicle/ layer — preserves the raw input, drafts a structured synthesis, and proposes updates to the chronicle entries the artifact informs.

Produces per-interview snapshots, theme rollups, opportunity-solution-tree updates, and a quote bank from discovery interview notes or transcripts. Use after the calls, when asked to synthesise, summarise, or find patterns across interviews. To plan the interview beforehand use discovery-interview-prep; to file a raw transcript into the chronicle use ingest.

Use to articulate why a customer hires a product — the functional, emotional, and social job they're trying to get done. Anchored in Clayton Christensen's JTBD theory.

Opens a conversational kickoff. Briefly explains what Hornero offers, invites the user to describe what is on their plate in natural language, and routes via the dispatcher with no pre-set menu of options.

Use when a PM runs the weekly triage pass on a Linear project — re-confirm strategy alignment first, then surface stale work, then trim the cycle scope, so the team's next cycle reflects current strategy with no hidden blockers.

Scans the project's memory/decisions/ directory for prior decisions whose scope, affected areas, or roles overlap the current work, and surfaces the relevant entries before fresh analysis begins. Use before reviewing, proposing, or producing a non-trivial artifact in a project that keeps lightweight decision files. If the project has a chronicle/ layer use chronicle-recall instead, which queries schema-validated entries.

Use to draft or critique quarterly OKRs — Objectives + Key Results — grounded in Christina Wodtke's radical-focus methodology.

Use when a PM needs to connect a desired outcome to the customer opportunities that could move it, map solution hypotheses against each opportunity, and select an experiment without collapsing the tree into a flat feature list.

Invokes the PM as cross-functional orchestrator. Frames the work as a customer job, picks the role mix, dispatches roles in parallel, and synthesizes a decision recommendation.

Use when a team needs to scan the external macro environment — Political, Economic, Social, Technological, Environmental, Legal — for forces that could shift strategy, inform a market-entry decision, or feed a strategy-kernel diagnosis.

Use to draft or critique a positioning statement using Geoffrey Moore's template, sharpened by April Dunford's framing.

Produces or critiques a PRD that anchors a feature in a problem statement, exposes solution-shaped thinking before it ossifies, pairs every success metric with a counter-metric, and names what is deliberately out of scope. Use when asked to write, review, or tighten a PRD or product spec. If the problem is not yet framed use problem-framing-canvas first; for a lightweight pre-implementation problem statement before code use spec-the-problem.

Use before a meeting, 1:1, design review, vendor call, or QBR — pulls the matching type: stakeholder entry, recent type: decision and open type: hypothesis entries in their scope, and recent ingestion notes, then renders a one-page brief with open asks, likely points of disagreement, and suggested questions.

Use when a team needs to write an Amazon-style working-backwards press release as a forcing function for problem clarity, customer-facing language, and the "is this worth building?" test before any spec is written.

Use to triage a backlog or set of options into Do-Now / Defer / Discard — applying LNO, ICE-with-confidence, and the four-risks lens to produce a ranked list with explicit reasoning and rank-change conditions.

Produces a structured problem statement from the customer's side: who has the problem, what triggers it, what happens if it is unsolved, what alternatives exist, and what success looks like for them. Use before drafting a PRD or committing to a solution. After the frame is settled use prd for the spec; for a quick code-facing problem and constraints pass before implementation use spec-the-problem.

Use when preparing for or critiquing a product-sense interview answer — to structure a spoken "design X" or "improve Y" response that sounds reasoned, not rehearsed.

Use when a team needs a lightweight, hypothesis-shaped persona to scope a problem or align on a target user before research is available — not a research-backed persona.

Cross-system read primitive over the team's issue tracker — Linear or Jira. Resolves an intent ("active blockers on the export project", "issues changed in the last 7 days by the auth team", "open enterprise-tier bugs") to a structured result without forcing the caller to learn the underlying query language.

Cross-system read primitive over the team's analytics warehouse — BigQuery or Snowflake. Three call shapes (schema discovery, dry-run cost estimate, exploratory SQL with a kill-cost guardrail) with stable result formats and a hard upper bound on bytes scanned per call.

Use when the PM needs to map the full product growth system — funnel, activation, retention, expansion, churn — as a connected graph, identify the bottleneck stage, and surface which levers move the system vs which produce only local lift.

Use weekly (typically Fridays) — runs six chronicle-health checks against the chronicle/entries/ corpus, proposes fixes for the unambiguous findings, and surfaces the rest for a human call. Pairs with /extract-lesson (which tends the harness) the way /review tends the chronicle.

Invokes a single Hornero role for a focused critique or artifact. Usage: /role <role-name> <task>. Available role: product-manager.

Explains and critiques SaaS unit-economics and capital-efficiency metrics (Rule of 40, burn multiple, gross margin, contribution margin, FCF), with the formula, healthy range, and common misread for each. Use when asked how efficiently the business turns spend into growth or profit. For ARR, NRR, GRR, CAC payback and expansion use saas-revenue-growth-metrics; for a one-page lookup of all ~30 metrics use finance-metrics-quickref; for a combined health read use business-health-diagnostic.

Explains and critiques SaaS growth and retention metrics (ARR, NRR, GRR, CAC payback, magic number, expansion levers), with the formula, healthy range, and common misread for each. Use when asked to read, sanity-check, or challenge top-of-funnel and retention numbers. For unit economics and capital efficiency (Rule of 40, burn multiple, margins, FCF) use saas-economics-efficiency-metrics; for a one-page lookup of all ~30 metrics use finance-metrics-quickref; for a combined health read use business-health-diagnostic.

Forces a problem statement, constraints, and success criteria pass before any non-trivial implementation, so code is not written against an unexamined request. Use at the start of an engineering task, bug fix, or refactor with unclear scope. It is deliberately short and code-facing: for a customer-centred problem frame ahead of a PRD use problem-framing-canvas, and for the full product spec use prd.

Use after any meaningful touchpoint with a stakeholder (1:1, ad-hoc Slack, meeting outcome) — appends to the relevant section on their type: stakeholder chronicle entry (Open asks, Concerns, Last contact, Decision affinity) via chronicle.sh append, keeping the entry current without forcing a full supersede.

Produces or critiques a strategy artifact in Rumelt's kernel shape: the central obstacle (diagnosis), the posture for overcoming it (guiding policy), and coherent reinforcing actions rather than an independent initiative list. Use when asked to write, review, or sharpen a product strategy. To make the bets inside a strategy explicit, with assumptions, evidence weight, and kill criteria, use bet-portfolio.

Produces a metric stack for a feature, launch or quarter: one primary metric, a counter-metric against gaming, guardrails for adjacent health, and leading indicators; also critiques an existing metric definition or dashboard metric for missing counter-metrics, unscoped populations and gaming surfaces. Use before instrumentation to define what good looks like, or when asked to review, challenge or tighten a metric definition. For finance metrics (ARR, NRR, Rule of 40, burn multiple) use saas-revenue-growth-metrics or saas-economics-efficiency-metrics; to draft OKRs use okrs.

Use when a PM needs to size a market — Total Addressable / Serviceable Available / Serviceable Obtainable — defensibly enough to defend a strategy, fund a bet, or shape a board claim, with top-down + bottom-up triangulation.

Use when a team needs to convert a discovery insight or product hypothesis into a release-shaped plan — Patton's two-dimensional story map with a narrative spine, walking skeleton, and release slices — that connects PM altitude to delivery sequencing.

Use when a PM needs to plan or run a focused, aligned product workshop — pre-work, facilitation modalities, time-boxing, decision-capture, and follow-through — rather than convening yet another open-ended discussion.