[aw] Evergreen failed

[github-actions]

Workflow Failure

Workflow: Evergreen
Branch: main
Run: https://github.com/githubnext/tsb/actions/runs/29052774215

Warning

Lock File Out of Sync: The workflow could not start because its compiled lock file no longer matches the source markdown.

This means the workflow's .md file was edited but gh aw compile was not run afterwards to regenerate the corresponding .lock.yml file. The agent is prevented from running against a stale configuration to avoid unexpected behaviour.

To fix, recompile the workflow:

gh aw compile

Then commit and push the updated .lock.yml file.

More ways to recompile

Using the gh-aw MCP server (if configured):

{ "tool": "compile", "arguments": { "validate": true } }

Recompile all workflows at once:

gh aw compile --all

Verify the result:

gh aw compile --validate

How to investigate the mismatch

The workflow run logs contain a verbose debug pass that shows exactly what was hashed. Search the Check workflow lock file step logs for lines starting with [hash-debug] to see:

• The raw frontmatter text that was used as input
• Any imported files that were included in the hash
• The canonical JSON that was fed to SHA-256
• The resulting hash value

This makes it easy to spot accidental whitespace changes, encoding differences, or import path drift.

This mismatch can also happen on a fresh install (even without any manual edits) if gh aw add <path>@<ref> could not resolve the ref to an exact commit SHA during installation. In that case, rerun the add command with an exact SHA (or retry when API/rate-limit conditions recover), then recompile:

gh aw add <path>@<exact-sha>
gh aw compile

How to disable this check

[!CAUTION]
Disabling this check means the agent can run against an out-of-date compiled workflow. Only disable it if you have an alternative mechanism to keep lock files in sync.

Set stale-check: false in the on: section of your workflow frontmatter:

on:
  issues:
    types: [opened]
  stale-check: false

After editing, recompile the workflow: gh aw compile

Warning

Daily Workflow AIC Guardrail Exceeded: The agent was not started because this workflow has already consumed the configured 24-hour AI Credits budget.

• 24h AIC usage: 10.2K AI Credits
• Configured threshold: 10K AI Credits

The agent will resume automatically once the 24-hour rolling window resets. No action is required if the current limit is appropriate for your usage.

_{AIC values are approximate. Consult the billing dashboards for accurate usage and charges.}

How to raise the daily limit

Set max-daily-ai-credits in your workflow frontmatter to a higher value, then recompile:

max-daily-ai-credits: 20K

Common suffix shorthands: K = thousands, M = millions (e.g. 2M = 2,000,000).

After editing the workflow source file, regenerate the compiled lock file:

gh aw compile

Commit and push the updated .lock.yml file.

[!NOTE]
Raising the limit increases the number of AI inference calls the workflow can make
per 24-hour window. Review your Copilot or model provider billing
before significantly increasing the threshold (for example, before doubling the current
value or setting it far above expected usage).

What is the daily AI Credits guardrail?

The max-daily-ai-credits frontmatter option sets a per-workflow spending cap measured in AI Credits across the 24-hour window before the current run. The cap is scoped to the repository and workflow — it aggregates usage across all runs of this workflow regardless of who triggered them.

When the aggregated AI Credits usage across all completed runs of this workflow in the last 24 hours exceeds the threshold, the activation job sets the daily_effective_workflow_exceeded output to true and the agent job is skipped for that run. The conclusion job still runs and creates this report.

The guardrail is evaluated at activation time, not retrospectively, so a single very large run that pushes usage over the threshold only blocks subsequent runs in the same window — it does not cancel a run that is already in progress.

How to disable this guardrail

[!CAUTION]
Disabling this guardrail removes the per-workflow spending cap. Only disable it if you have
an alternative mechanism for controlling AI cost usage or if the workflow is intentionally
uncapped.

Set max-daily-ai-credits: -1 in the workflow frontmatter to explicitly disable the guardrail, then recompile:

max-daily-ai-credits: -1

gh aw compile

Alternatively, remove the max-daily-ai-credits key entirely to fall back to the enterprise-wide default (if one is configured) or to run with no per-workflow cap.

Action Required

Assign this issue to an agent to debug and fix the issue.

Debug with any coding agent

Use this prompt with any coding agent (GitHub Copilot, Claude, Gemini, etc.):

Debug the agentic workflow failure using https://raw.githubusercontent.com/github/gh-aw/main/debug.md

The failed workflow run is at https://github.com/githubnext/tsb/actions/runs/29052774215

Manually invoke the agent

Debug this workflow failure using your favorite Agent CLI and the agentic-workflows prompt.

• Start your agent
• Load the agentic-workflows skill from .github/skills/agentic-workflows/SKILL.md or https://github.com/github/gh-aw/blob/main/.github/skills/agentic-workflows/SKILL.md
• Type debug the agentic workflow evergreen failure in https://github.com/githubnext/tsb/actions/runs/29052774215

Tip

Stop reporting this workflow as a failure

To stop a workflow from creating failure issues, set report-failure-as-issue: false in its frontmatter:

safe-outputs:
  report-failure-as-issue: false

Generated from Evergreen · ◷

• expires on Jul 16, 2026, 9:54 PM UTC