Skip to content

v4.7.0 — PM becomes a production toolkit (54 skills, ~30k lines added)

Choose a tag to compare

@borghei borghei released this 22 May 10:20
· 22 commits to main since this release

TL;DR

Project Management is now the deepest domain in Claude Skills. 54 skills spanning discovery → delivery → career growth, with worked examples for every skill, anti-pattern libraries, runnable pipelines, live data adapters, MCP tool integration, and a deterministic output eval harness.

~30,000 lines added since v4.2.0, across 9 atomic commits and 12+ background-agent passes.

What's new since v4.2.0

Domain expansion (v4.3.0 + v4.4.0)

  • PM grew from 25 → 54 skills (+29 new). Added modern AI/growth PM (ai-feature-prd, activation-funnel, feature-flag-strategy), strategy (product-vision, quarterly-planning), execution gaps (post-mortem, customer-feedback-triage, pricing-prd), career growth (pm-interview-prep, pm-career-ladder, pm-onboarding, pm-1on1s), integrations (linear-expert, notion-pm, productboard-expert), and discovery (customer-interview-script, value-proposition-canvas, jtbd-workshop, interview-synthesis).
  • Library total: 245 → 294 skills across 15 domains.
  • 4 new PM-focused bundles in bundles.json: pm-starter-pack, pm-scrum-master, pm-discovery-coach, pm-career-track.

Tier 1 of the depth investment (v4.5.0)

  • 54 worked examples, one per skill (~13k lines). Each has a realistic fictional scenario (Acme Analytics, Wayfinder, Northwind, Helix Platform, Pylon), the workflow applied, and the FULL generated artifact PMs can copy.
  • 3 live data adapters at tools/adapters/ — pull from Jira / Linear / Notion APIs and emit JSON in the shapes PM Python tools expect. Stdlib only, env-var auth.
  • 15 MCP tools registered in scripts/mcp_server.py. Each PM skill callable from any AI conversation: pm_create_prd, pm_status_update, pm_funnel_analyze, pm_flow_metrics, pm_dependency_map, pm_feedback_triage, pm_nsm_tree, pm_refinement_score, pm_interview_synthesize, pm_prioritize, pm_okr_validate, pm_roadmap_transform, pm_pre_mortem, pm_release_notes, pm_stakeholder_map.

Tier 2 of the depth investment (v4.6.0)

  • 54 red-flag libraries, one per skill (~12k lines). Each lists 10-12 concrete anti-patterns with paired bad/good quoted artifact snippets and a "How to catch it" check question. Anchored to canonical authors (Wodtke, Wake, Lawrence, Vacanti, Allspaw, Dekker, Klein, Klement, Ellis, McClure, Karpathy, Anthropic RSP, Ramanujam, Westendorp, etc.).
  • 5 runnable pipelines at pipelines/ — chain 3-7 PM skills end-to-end with one command: feature-end-to-end, weekly-cadence, customer-discovery, post-mortem-flow, launch-coordination. Each supports --demo, --input, --format markdown|json, --output <dir>.

Tier 3 of the depth investment (v4.7.0)

  • Output eval harness at evals/ — deterministic scoring of any markdown artifact, no LLM needed.
    • engine.py — 13 supported check types (regex, section_present, keyword_none, has_table, etc.)
    • run.py — runner with --all, --skill, --artifact, --threshold, --format flags
    • 12 rubrics with 190+ criteria total, each anchored to a specific red-flag or success-criterion item
  • Smoke-test: 12/12 worked examples pass at threshold 70. Average score 96.4.

Infrastructure fixes

  • scripts/build_manifest.py now recurses into container folders (discovery/, execution/, career/). Previously the registry undercounted PM by ~30 nested skills.
  • .claude/settings.json hooks use $CLAUDE_PROJECT_DIR instead of relative paths — works regardless of CWD.
  • .github/workflows/sync-codex-skills.yml retries push on concurrent-bot race condition (rebase + retry up to 3 times). Fixes a fragile-but-non-blocking workflow failure.

Website (borghei.github.io/Claude-Skills)

  • New dedicated PM landing page at /skills/project-management/.
  • Front-page hero refreshed: 266 → 294 skills, 17 → 15 domains, "cs-* Agents" → "Expert Agents". New 18-compliance-frameworks stat.
  • PM domain card highlighted with border + new tags (ai-prd, linear, notion, productboard).
  • Section heading rewrites: "Key Features" → "How it works"; "Cross-Domain Personas" → "Pre-built expert identities"; "Starter Bundles" → "Ready-made kits by role"; "Platform Support" → "Works with your AI assistant".
  • Install section now leads with Try it now (chat) + Save to my AI; npx demoted to "For developers".
  • All em dashes replaced with context-appropriate punctuation (period / comma / colon / parens) across README and front page.

Repository metadata

  • GitHub description updated to highlight PM and regulated industries.
  • Topics refreshed: added project-management, product-management, linear, notion, jira, agile; removed lower-value tags.

Branch hygiene

  • Two fully-merged stale feature branches removed (feature/product-pm-skills-upgrade, feature/skill-pattern-adoption).
  • Repo is one branch only: main.

Final PM domain layout

Layer Count
Skills 54
Worked examples (per-skill) 54
Red-flag libraries (per-skill) 54
Python tools 9+
Live data adapters 3 (Jira / Linear / Notion)
MCP tools 15
Runnable pipelines 5
Eval rubrics 12
Career sub-domain (new) 4 skills
Integration skills 3 (Linear, Notion, Productboard)

Breaking changes

None. Fully additive.

Upgrade path

npx @borghei/claude-skills add pm-starter-pack

Or browse the PM domain on the live site.