Problem Description
When calling the logs tool with a workflow_name filter, the returned run objects are missing several fields that are present when calling without a filter. This creates inconsistent data depending on the query type.
Tool
Steps to Reproduce
- Call
logs without any filters (unfiltered path)
- Call
logs with workflow_name: "bot-detection" (filtered path)
- Compare the run object fields
Expected Behavior
Run objects should have a consistent set of fields regardless of whether a workflow_name filter is applied.
Actual Behavior
Filtered runs (by workflow_name) are missing these fields:
agent (e.g., "codex")
engine (e.g., "Codex")
engine_id (e.g., "codex")
actor (e.g., "pelikhan")
organization (e.g., "github")
repository (e.g., "github/gh-aw")
ref (e.g., "refs/heads/main")
sha (full commit SHA)
run_attempt
event_name
Meanwhile, filtered runs have additional fields not in unfiltered:
avg_time_between_turns
conclusion
effective_tokens
temporary_id_map_status
turns
This suggests the two paths use different data sources and enrichment logic that aren't reconciled.
Environment
- Repository: github/gh-aw
- Workflow Run: §26436212879
- Date: 2026-05-26
Impact
- Severity: Medium
- Frequency: Always (reproducible)
- Workaround: None — callers must handle missing fields defensively
Diagnostic Evidence
Unfiltered run object includes:
{
"run_id": 26436319891,
"agent": "codex",
"engine": "Codex",
"engine_id": "codex",
"actor": "pelikhan",
"organization": "github",
"repository": "github/gh-aw",
"ref": "refs/heads/main",
"sha": "ccf77c4da6b6fa1af0826b4542c3259213fbb402"
}
Filtered run object (same workflow, by workflow_name: "bot-detection") is missing all the above fields.
Additional Context
Tested during daily CLI tools exploratory testing session (run §26436212879).
Generated by 🧪 Daily CLI Tools Exploratory Tester · sonnet46 2.2M · ◷
Problem Description
When calling the
logstool with aworkflow_namefilter, the returned run objects are missing several fields that are present when calling without a filter. This creates inconsistent data depending on the query type.Tool
logsSteps to Reproduce
logswithout any filters (unfiltered path)logswithworkflow_name: "bot-detection"(filtered path)Expected Behavior
Run objects should have a consistent set of fields regardless of whether a
workflow_namefilter is applied.Actual Behavior
Filtered runs (by
workflow_name) are missing these fields:agent(e.g.,"codex")engine(e.g.,"Codex")engine_id(e.g.,"codex")actor(e.g.,"pelikhan")organization(e.g.,"github")repository(e.g.,"github/gh-aw")ref(e.g.,"refs/heads/main")sha(full commit SHA)run_attemptevent_nameMeanwhile, filtered runs have additional fields not in unfiltered:
avg_time_between_turnsconclusioneffective_tokenstemporary_id_map_statusturnsThis suggests the two paths use different data sources and enrichment logic that aren't reconciled.
Environment
Impact
Diagnostic Evidence
Unfiltered run object includes:
{ "run_id": 26436319891, "agent": "codex", "engine": "Codex", "engine_id": "codex", "actor": "pelikhan", "organization": "github", "repository": "github/gh-aw", "ref": "refs/heads/main", "sha": "ccf77c4da6b6fa1af0826b4542c3259213fbb402" }Filtered run object (same workflow, by
workflow_name: "bot-detection") is missing all the above fields.Additional Context
Tested during daily CLI tools exploratory testing session (run §26436212879).