Skip to content

Bound AgentRx optimizer input to avoid prompt exhaustion #67133

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
pelikhan merged 2 commits into from
Oct 9, 2026
Merged

Bound AgentRx optimizer input to avoid prompt exhaustion #67133

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
a495d84
Initial plan
Copilot Oct 9, 2026
76543d2
Bound AgentRx log analysis input
Copilot Oct 9, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions .github/workflows/daily-agentrx-trace-optimizer.lock.yml
View file
Open in desktop

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

8 changes: 5 additions & 3 deletions .github/workflows/daily-agentrx-trace-optimizer.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -102,13 +102,13 @@ Focus on:
- Use `logs` to download parsed logs for recent runs, specifying `artifacts: ["agent"]` to include agent telemetry (turns, token usage, stdout) needed for AgentRx trajectory analysis.
**`logs` precondition rules (follow strictly):**
- Always include `workflow_name` — never call `logs` without it; an unfiltered scan will time out.
- Cap retries at **2 attempts** per workflow. If both return empty or time out, stop retrying and fall back to `audit` using any known `run_id` (from `status` or prior context).
- Query at most **3 workflows**, with `count: 10` and `start_date: "-1d"` for each. Cap retries at **2 attempts** per workflow; if both return empty or time out, stop retrying and fall back to `audit` using any known `run_id` (from `status` or prior context). Never widen the time window or run count.
- If `logs` returns `total_runs=0` for a workflow with confirmed completed runs, treat this as a tool-health failure. Surface it via `missing_tool` and proceed with `audit`-only data; do not vary parameters and retry further.
```json
{
"workflow_name": "<workflow-id>",
"count": 50,
"start_date": "-2d",
"count": 10,
"start_date": "-1d",
"artifacts": ["agent"]
}
```
Expand Down Expand Up @@ -140,6 +140,8 @@ run_data_path: /tmp/gh-aw/agent/agentrx/mcp-runs.json
Do not invoke `trajectory-builder` in this variant.
{{/if}}

Limit the trajectory to **10 runs total** across all sources. Prioritize failed runs, then the highest-latency runs, and omit raw log text.

### 2) Run AgentRx pipeline

Run the pipeline in stages and preserve outputs under `/tmp/gh-aw/agent/agentrx/runs/<run_name>/`:
Expand Down
Loading