Harden API consumption log collection against continuation timeouts

[Copilot]

The API consumption report could publish a partial 24h window when a continuation call timed out mid-collection, dropping already-fetched coverage and skewing day-over-day comparisons. This change hardens the workflow’s log-collection instructions so continuation failures degrade to a preserved partial dataset instead of silently losing progress.

• Collection resilience

  • Treat /tmp/gh-aw/aw-mcp/logs/ as the authoritative collected dataset across pagination calls.
  • Keep already-downloaded run directories when a later continuation call fails, instead of treating the failed page as an all-or-nothing boundary.

• Bounded continuation retries

  • Retry the same continuation request up to 2 additional times on transient failures (timeout, ECONNREFUSED, transport/tool errors).
  • Require short backoff between retries without mutating returned pagination parameters.

• Pagination discipline

  • Re-read the returned continuation field after each successful follow-up call.
  • Continue to forbid inventing before_run_id or using ad hoc pagination/count tuning outside the MCP response.

• Report clarity on degraded runs

  • Instruct the workflow to explicitly note when continuation retries were exhausted and the final dataset may still be partial.

- Treat the run directories already present under `/tmp/gh-aw/aw-mcp/logs/` as the authoritative collected dataset.
- If a continuation call times out ... retry that same continuation call up to **2** more times with short backoff.
- If the continuation call still fails after those bounded retries, stop collecting and proceed with the logs already downloaded.

[Copilot]

Pull request overview

Note

Copilot was unable to run its full agentic suite in this review.

Updates the API consumption report workflow instructions to make log pagination/continuation more resilient and to preserve partial results when continuation fails.

Changes:

• Treat on-disk /tmp/gh-aw/aw-mcp/logs/ directories as the authoritative collected dataset during pagination.
• Add bounded retries with backoff for transient continuation failures and clarify how to proceed with partial data.
• Regenerate the workflow lock/metadata to reflect the updated markdown content.

Show a summary per file

File	Description
.github/workflows/api-consumption-report.md	Refines log collection/pagination guidance (authoritative on-disk dataset + bounded retries).
.github/workflows/api-consumption-report.lock.yml	Updates generated metadata hashes corresponding to the markdown changes.

Review details

Tip

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

• Files reviewed: 2/2 changed files
• Comments generated: 2
• Review effort level: Low