fix(schema-consistency-checker): replace max-turns with max-effective-tokens 20M and reduce turn waste

[Copilot]

The schema-consistency-checker has been failing error_max_turns on 5/7 days. The one successful run (May 6) consumed 101 turns — 68% over the 60-turn cap.

Changes

• Replaced max-turns with harness.budget.max-effective-tokens: 20000000 — switches from a turn-count budget to a 20 M token budget, giving the workflow reliable headroom without an arbitrary turn ceiling
• Merged Step 0 + Step 1 into a single bash command — schema-diff and strategy-cache are now read in one tool call instead of two, saving a turn on every run:

  echo "=== SCHEMA DIFF ===" && cat /tmp/gh-aw/agent/schema-diff.json && \
  echo "=== STRATEGIES ===" && \
  ([ -f /tmp/gh-aw/cache-memory/strategies.json ] && cat /tmp/gh-aw/cache-memory/strategies.json || echo "No strategies cached yet")

• Added "Turn Budget" section — explicit guidance to batch bash calls, stop at 3+ findings, prioritize field_gaps (pre-computed, zero discovery cost), and skip low-signal categories rather than exhaustively covering all four areas every run
• Recompiled lock file — removes --max-turns CLI flag and GH_AW_MAX_TURNS env var; updates GH_AW_MAX_EFFECTIVE_TOKENS to "20000000"

[!WARNING]

[pelikhan]

@copilot remove max-turns and replace with max-effective-tokens 20M

[Copilot]

@copilot remove max-turns and replace with max-effective-tokens 20M

Done — max-turns removed and replaced with harness.budget.max-effective-tokens: 20000000 in both the workflow definition and the compiled lock file (commit fix(schema-consistency-checker): replace max-turns with max-effective-tokens 20M).

[Copilot]

Pull request overview

Updates the schema-consistency-checker workflow configuration to reduce failures caused by hitting execution limits, primarily by shifting from a turn-based cap to an effective-token budget and by reducing wasted turns in the prompt instructions.

Changes:

• Swaps the workflow budgeting approach from engine.max-turns to an effective-token budget and adds guidance to conserve turns/tool calls.
• Merges the “read pre-computed data” and “load strategies” steps into a single batched bash command in the instructions.
• Recompiles the workflow lock file to remove --max-turns / GH_AW_MAX_TURNS and update the effective-token budget env var.

Show a summary per file

File	Description
.github/workflows/schema-consistency-checker.md	Updates workflow frontmatter and prompt instructions to change budgeting strategy and reduce tool-call waste.
.github/workflows/schema-consistency-checker.lock.yml	Recompiled lock to remove max-turns wiring and update effective-token budget configuration.

Copilot's findings

Tip

Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

• Files reviewed: 2/2 changed files
• Comments generated: 3