feat(skills): add process-impact playbook for change justification

[adriannoes]

Summary

• Add consulting skill pipefy-process-impact so agents can quantify whether a process change is worth it from context already in the conversation, without implementing it.
• Process-design and process-intelligence emit a 1–2 line impact blurb on material changes; the building router sends ROI / justification asks here.
• Follow-up contract: diagnosis reuses intelligence round 1 only; get_ai_agents takes the pipe UUID; cards_count is WIP, not lead time; a card sample is not weekly volume.

Test plan

• uv run python .github/workflows/scripts/lint_skill_refs.py (Wave 0: passed, 18 files)
• uv run pytest tests/test_process_impact_skill_catalog.py (Wave 0: 4 passed)
• uv run pytest -m "not integration"
• uv run ruff check . && uv run ruff format --check .
• Routing smoke (HIT vs MISS): 11/11 Pass
• Live read on the QA Analytics pipe (numeric id plus uuid hop for get_ai_agents)

Routing scenarios

HIT (load this skill):

• ROI / worth-it triage
• Analyze and "is it worth it?"
• Delete pipes / go back to Excel
• Lead time in days
• Measure hops / volume / AI agents (live)
• BRL plus headcount
• Treat a 50-card page as weekly volume (must refuse)

MISS (do not load this skill):

• Design a purchase process
• Just build the automation
• "What is process impact?"
• Diagnose bottlenecks with no ROI ask

Live traps (QA Analytics pipe)

• Numeric pipe id on get_ai_agents returns PERMISSION_DENIED. Resolve pipe.uuid via get_pipe first, then get_ai_agents repo_uuid=<pipe.uuid>.
• phases[].cards_count is live WIP / bottleneck inventory, not lead time.
• A card sample (page size) is not cases_per_week.

Docs / skills

• docs/parity.md updated when MCP ↔ CLI coverage changed
• Affected skills/ updated in this PR (or a paired PR)
• No docs/skills update needed

Catalog and plugin wiring land with the skill: skills/README.md, skills/AGENTS.md domain list, CHANGELOG.md Unreleased, .cursor-plugin/plugin.json. No MCP/CLI tools added, so parity is unchanged. Root README.md and docs/ keep pointing at the skills catalog rather than listing each skill.

Legal / contributions

• Commits include DCO sign-off (git commit -s)
• Regulated-domain skills include COMPLIANCE.md when applicable