feat(ce-plan): extract conditional blocks to references/ for token savings

Extract three large conditional/late-sequence blocks from ce:plan SKILL.md
into on-demand reference files, reducing the base skill from 1,023 lines
(~15,525 tokens) to 685 lines (~9,971 tokens) — a 36% reduction.

For orchestration-heavy skills like ce:plan with 20-40+ tool/agent calls
per session, unused content carried in every message exchange compounds
significantly. This saves ~130,000-167,000 context tokens per planning
session depending on which code paths execute.

Extracted files:
- references/deepening-workflow.md (5.3.3-5.3.7, ~3,808 tok) — loaded
  only when the deepening gate passes
- references/visual-communication.md (4.4, ~823 tok) — loaded only when
  plan has complex visual-worthy content patterns
- references/plan-handoff.md (5.3.8-5.4, ~1,304 tok) — loaded only at
  end of workflow for document review and post-generation options

Also adds a "Conditional and Late-Sequence Extraction" rule to the plugin
AGENTS.md to guide future skill optimization.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: tmchow

plugins/compound-engineering/AGENTS.md

@@ -93,6 +93,10 @@ When adding or modifying skills, verify compliance with the skill spec:
   This resolves relative to the SKILL.md and substitutes content before the model sees it. If a file is over ~150 lines, prefer a backtick path even if it is always needed
 - [ ] For files the agent needs to *execute* (scripts, shell templates), always use backtick paths -- `@` would inline the script as text content instead of keeping it as an executable file
 
+### Conditional and Late-Sequence Extraction
+
+Skill content loaded at trigger time is carried in every subsequent message — every tool call, agent dispatch, and response. This carrying cost compounds across the session. For skills that orchestrate many tool or agent calls, extract blocks to `references/` when they are conditional (only execute under specific conditions) or late-sequence (only needed after many prior calls) and represent a meaningful share of the skill (~20%+). The more tool/agent calls a skill makes, the more aggressively to extract. Replace extracted blocks with a 1-3 line stub stating the condition and a backtick path reference (e.g., "Read `references/deepening-workflow.md`"). Never use `@` for extracted blocks — it inlines content at load time, defeating the extraction.
+
 ### Writing Style
 
 - [ ] Use imperative/infinitive form (verb-first instructions)