9d55d531a4
- quick.md Step 5.6: commit PLAN.md to base branch before worktree executor spawn when USE_WORKTREES is active, preventing CC #36182 path-resolution drift that caused silent writes to main repo instead of worktree - reapply-patches.md Option A: replace first-add commit heuristic with pristine_hashes SHA-256 matching from backup-meta.json so baseline detection works correctly on multi-cycle repos; first-add fallback kept for older installers without pristine_hashes - CONFIGURATION.md: move security_enforcement/security_asvs_level/security_block_on to workflow.* (matches templates/config.json and workflow readers); rename context_profile → context (matches VALID_CONFIG_KEYS in config.cjs); add planning.sub_repos to schema example - universal-anti-patterns.md + context-budget.md: fix context_window_tokens → context_window (the actual key name in config.cjs) Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
3.3 KiB
3.3 KiB
Context Budget Rules
Standard rules for keeping orchestrator context lean. Reference this in workflows that spawn subagents or read significant content.
See also: references/universal-anti-patterns.md for the complete set of universal rules.
Universal Rules
Every workflow that spawns agents or reads significant content must follow these rules:
- Never read agent definition files (
agents/*.md) --subagent_typeauto-loads them - Never inline large files into subagent prompts -- tell agents to read files from disk instead
- Read depth scales with context window -- check
context_windowin.planning/config.json:- At < 500000 tokens (default 200k): read only frontmatter, status fields, or summaries. Never read full SUMMARY.md, VERIFICATION.md, or RESEARCH.md bodies.
- At >= 500000 tokens (1M model): MAY read full subagent output bodies when the content is needed for inline presentation or decision-making. Still avoid unnecessary reads.
- Delegate heavy work to subagents -- the orchestrator routes, it doesn't execute
- Proactive warning: If you've already consumed significant context (large file reads, multiple subagent results), warn the user: "Context budget is getting heavy. Consider checkpointing progress."
Read Depth by Context Window
| Context Window | Subagent Output Reading | SUMMARY.md | VERIFICATION.md | PLAN.md (other phases) |
|---|---|---|---|---|
| < 500k (200k model) | Frontmatter only | Frontmatter only | Frontmatter only | Current phase only |
| >= 500k (1M model) | Full body permitted | Full body permitted | Full body permitted | Current phase only |
How to check: Read .planning/config.json and inspect context_window. If the field is absent, treat as 200k (conservative default).
Context Degradation Tiers
Monitor context usage and adjust behavior accordingly:
| Tier | Usage | Behavior |
|---|---|---|
| PEAK | 0-30% | Full operations. Read bodies, spawn multiple agents, inline results. |
| GOOD | 30-50% | Normal operations. Prefer frontmatter reads, delegate aggressively. |
| DEGRADING | 50-70% | Economize. Frontmatter-only reads, minimal inlining, warn user about budget. |
| POOR | 70%+ | Emergency mode. Checkpoint progress immediately. No new reads unless critical. |
Context Degradation Warning Signs
Quality degrades gradually before panic thresholds fire. Watch for these early signals:
- Silent partial completion -- agent claims task is done but implementation is incomplete. Self-check catches file existence but not semantic completeness. Always verify agent output meets the plan's must_haves, not just that files exist.
- Increasing vagueness -- agent starts using phrases like "appropriate handling" or "standard patterns" instead of specific code. This indicates context pressure even before budget warnings fire.
- Skipped steps -- agent omits protocol steps it would normally follow. If an agent's success criteria has 8 items but it only reports 5, suspect context pressure.
When delegating to agents, the orchestrator cannot verify semantic correctness of agent output -- only structural completeness. This is a fundamental limitation. Mitigate with must_haves.truths and spot-check verification.