Problem Description
The MCP compile tool reports successful compilation of copilot-agent-analysis, but a subsequent MCP status query still reports compiled: "No". This breaks the status → compile → status freshness check and can cause agents to repeatedly recompile an apparently stale workflow.
Command/Tool
- Tools:
compile and status, through the configured agenticworkflows MCP bridge.
- Repository: github/gh-aw.
- Date: 2026-10-05.
- Test run: §37268376949.
- gh-aw version: not supplied by
status.
Steps to Reproduce
- Query
agenticworkflows status '{"pattern":"copilot-agent-analysis"}'.
- Compile with
agenticworkflows compile '{"workflows":["copilot-agent-analysis"],"strict":true}'.
- Query the same status again after compilation completes.
The same stale status was observed after a successful full-repository compilation. No workflow sources were edited during these tests.
Expected Behavior
Successful compilation should leave the workflow up to date (compiled: "Yes"), or return a diagnostic explaining why the generated output remains stale.
Actual Behavior
Compilation returns valid: true, empty errors and warnings, and a compiled lock-file path. Status still returns compiled: "No". The lock file exists locally.
Impact
- Severity: Medium.
- Frequency: observed after both bulk and targeted compilation in this session; broader reproducibility is unverified.
- Workaround: none established through the tested MCP interface.
Diagnostics and scope
Targeted compilation returned:
{
"workflow": "copilot-agent-analysis.md",
"valid": true,
"errors": [],
"warnings": [],
"compiled_file": "/home/runner/work/gh-aw/gh-aw/.github/workflows/copilot-agent-analysis.lock.yml"
}The subsequent status response included:
{
"workflow": "copilot-agent-analysis",
"engine_id": "claude",
"compiled": "No",
"status": "active",
"dependencies": [
"shared/copilot-pr-analysis-base.md",
"shared/daily-audit-base.md",
"shared/otlp.md",
"shared/repo-memory-standard.md"
]
}Initial discovery found 320 workflows, with 319 reporting compiled: "Yes" and this workflow reporting "No". Bulk compile returned 320 valid workflows and no warnings in 28.5 seconds. The targeted recompile completed in 0.31 seconds; the follow-up status completed in 0.60 seconds.
Source inspection shows isCompiledUpToDateWithCache in pkg/cli/status_command.go compares the computed frontmatter hash with lock metadata. The root cause of the disagreement has not been established; this report does not assert whether compilation or freshness detection is responsible.
Duplicate search was attempted, but matching issue contents were withheld by the integration's integrity policy. No attempt was made to bypass that restriction.
Next Actions
Investigate why the compiler and status freshness check disagree for this workflow and add a regression check that successful compilation makes the subsequent status current.
References:
Generated by 🧪 Daily Cli Tools Tester · codex · gpt60 · 189.4 AIC · ⌖ 37 AIC · ⊞ 19.6K · ◷
Problem Description
The MCP
compiletool reports successful compilation ofcopilot-agent-analysis, but a subsequent MCPstatusquery still reportscompiled: "No". This breaks the status → compile → status freshness check and can cause agents to repeatedly recompile an apparently stale workflow.Command/Tool
compileandstatus, through the configuredagenticworkflowsMCP bridge.status.Steps to Reproduce
agenticworkflows status '{"pattern":"copilot-agent-analysis"}'.agenticworkflows compile '{"workflows":["copilot-agent-analysis"],"strict":true}'.The same stale status was observed after a successful full-repository compilation. No workflow sources were edited during these tests.
Expected Behavior
Successful compilation should leave the workflow up to date (
compiled: "Yes"), or return a diagnostic explaining why the generated output remains stale.Actual Behavior
Compilation returns
valid: true, empty errors and warnings, and a compiled lock-file path. Status still returnscompiled: "No". The lock file exists locally.Impact
Diagnostics and scope
Targeted compilation returned:
{ "workflow": "copilot-agent-analysis.md", "valid": true, "errors": [], "warnings": [], "compiled_file": "/home/runner/work/gh-aw/gh-aw/.github/workflows/copilot-agent-analysis.lock.yml" }The subsequent status response included:
{ "workflow": "copilot-agent-analysis", "engine_id": "claude", "compiled": "No", "status": "active", "dependencies": [ "shared/copilot-pr-analysis-base.md", "shared/daily-audit-base.md", "shared/otlp.md", "shared/repo-memory-standard.md" ] }Initial discovery found 320 workflows, with 319 reporting
compiled: "Yes"and this workflow reporting"No". Bulk compile returned 320 valid workflows and no warnings in 28.5 seconds. The targeted recompile completed in 0.31 seconds; the follow-up status completed in 0.60 seconds.Source inspection shows
isCompiledUpToDateWithCacheinpkg/cli/status_command.gocompares the computed frontmatter hash with lock metadata. The root cause of the disagreement has not been established; this report does not assert whether compilation or freshness detection is responsible.Duplicate search was attempted, but matching issue contents were withheld by the integration's integrity policy. No attempt was made to bypass that restriction.
Next Actions
Investigate why the compiler and status freshness check disagree for this workflow and add a regression check that successful compilation makes the subsequent status current.
References: