Agent · PM 📋

E11 PM (skill: e11-agent-pm) — the product manager. User-interview-driven PRDs, epics and stories, the readiness gate, and mid-flight course correction. The PM also carries the planning-side stewardship the retired SM agent used to own.

What it produces

• A PRD — {planning_artifacts}/prds/prd-<project>-<date>/prd.md plus its addendum.md and .memlog.md audit trail (via the consolidated e11-prd).
• An epics + stories listing — {planning_artifacts}/epics.md, every story stamped with the @engineering11/* SDKs it touches.
• A readiness report — the gate before Phase 4.
• A sprint change proposal — {planning_artifacts}/sprint-change-proposal-<date>.md when course correction is needed.

When to use it

• After the brief: you need a PRD (or need an existing one updated or validated — all three are one skill now).
• After architecture: you need the epic/story breakdown.
• Before implementation: run the readiness check.
• Mid-implementation: something is materially off — course-correct instead of quietly editing artifacts.

How to invoke

Invoke e11-agent-pm; it greets and presents its menu:

Code	Dispatches to	Purpose
[PR]	e11-prd	Create / Update / Validate a PRD (one skill, three intents)
[CE]	e11-create-epics-and-stories	PRD + architecture → epics + stories with SDK stamps
[IR]	e11-check-implementation-readiness	Cross-artifact alignment gate before Phase 4
[CC]	e11-correct-course	Continue / update PRD / redo architecture / restart decision
[QK]	e11-knowledge	Pull an alloy doc directly

The consolidated PRD

[PR] replaces the earlier separate create/validate/edit workflows. It detects intent, then calibrates depth by stakes (hobby ≈ 2 pages, internal 5–8, launch as needed) and lets you pick the fast path (batched questions → full draft with [ASSUMPTION] tags) or the coaching path (section-by-section with pushback). During discovery it surveys the platform (skill:e11-knowledge list:sdks …) and writes a Platform Capability Map into the PRD — every MVP feature classified Covered / Composed / New with alloy:// citations. That map is what the Architect, the story stamps, and the review auditor all trace back to.

Example session

You: e11-agent-pm

PM: 📋 E11 PM. What are we shaping?

You: PR — PRD for the tenant usage dashboard. Internal tool.

PM: Stakes: internal — I'll aim for 5–8 pages. Fast path or coaching?

You: Fast.

PM: Brain-dump everything you know; I'll ask one batch of gap questions,
    then draft in full with [ASSUMPTION] tags.
    ...
    Platform capability survey: skill:e11-knowledge list:sdks
    - Usage metering: Covered (cites alloy://sources/sdks-nodejs/sdks/...)
    - Dashboard shell: Composed (2 SDKs)
    - Export scheduler: New → architecture follow-up
    Draft at planning-artifacts/prds/prd-usage-dashboard-2026-07-20/prd.md.

Gotchas

• Don't slip scope in through Update. Mid-implementation scope changes go through [CC] first, then the PRD update applies the agreed change.
• The finalize gate is real. The PRD's finalize step runs reviewer subagents (including one that enforces the capability map's alloy:// cites resolve against the pinned corpus via an alloy check-consumers sweep). Stakes calibrate how heavy it is - hobby projects may skip.
• Old names still work, once. e11-create-prd / e11-edit-prd / e11-validate-prd forward to e11-prd with the right intent, with a deprecation notice. See Upgrading.

Customizing

Override the persona, menu, PRD template (prd_template), output paths, reviewer roster (finalize_reviewers), or turn off the capability map (e11_platform_capability_map = false) in .foundry/custom/e11-prd.toml / .foundry/custom/e11-agent-pm.toml. See Customization.

Source

src/foundry/workflows/2-plan-workflows/e11-agent-pm/SKILL.md

See also: Phase 2 — Planning · Phase 3 — Solutioning · Knowledge layer.