githubnext / githubnext/gh-aw-wizard

[aw] Prompt Optimization reported incomplete result

Open
#252 0 comments 0 reactions 0 assignees View on GitHub
agentic-workflows
Dominant language
JavaScript
Stars
3
Forks
0
Avg merge
3h 8m
Merged PRs (30d)
244

Description

### Workflow Failure

**Workflow:** [Prompt Optimization](https://github.com/githubnext/gh-aw-wizard/blob/main/.github/workflows/prompt-optimization.md)
**Branch:** main
**Run:** https://github.com/githubnext/gh-aw-wizard/actions/runs/33684556510

> [!WARNING]
> **Task Could Not Be Completed**: The agent reported that the task could not be performed due to an infrastructure or tool failure.

**Reasons:**
- Ollama server was not running at (127.0.0.1/redacted) as the task claimed (connection refused, no ollama process, empty model cache). Started ollama serve myself, but pulling both required GGUF proxy models (bartowski/Qwen2.5-1.5B-Instruct-GGUF and unsloth/SmolLM2-360M-Instruct-GGUF) from huggingface.co failed with 403 Forbidden — the sandbox's squid proxy blocks hf.co/huggingface.co entirely (ERR_ACCESS_DENIED). No cached/pre-downloaded model weights exist anywhere in the environment. Without a reachable eval endpoint serving these models, no evaluation, scoring, or hill-climbing could be performed, so no candidate prompt could be measured or adopted.

Checked: no `ollama` process running initially (verified via ps aux); binary present at /opt/hostedtoolcache/ollama/0.33.2/x64/ollama, started manually with OLLAMA_MODELS=/tmp/gh-aw/agent/ollama-models — server came up healthy ("Ollama is running") but with 0 blobs/models loaded. `ollama pull hf.co/bartowski/Qwen2.5-1.5B-Instruct-GGUF:Q4_K_M` failed: "pull model manifest: Get (hf.co/redacted) Forbidden". Direct curl to (hf.co/redacted) and (huggingface.co/redacted) through the configured HTTPS_PROXY (172.30.0.10:3128) returns HTTP 403 from squid (X-Squid-Error: ERR_ACCESS_DENIED). No .gguf files or Ollama manifest/blob directories found anywhere on disk. Confirmed no open PR titled "Prompt optimization:" existed, so this was a fresh optimization attempt, not a review-existing-PR path. Reverted the one incidental change to .github/skills/optimize-scenario-prompt/SKILL.md (front-matter reordering from skill loading) so the working tree is clean; no other repository files were modified, and no PR, review comment, or review was created.

This is a structured incompletion signal (`report_incomplete`), not a real task outcome. Any other safe outputs emitted alongside this signal (e.g., comments) describe the failure state, not a completed review or action.

### 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/gh-aw-wizard/actions/runs/33684556510
````

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
- Type `debug the agentic workflow prompt-optimization failure in https://github.com/githubnext/gh-aw-wizard/actions/runs/33684556510`

> [!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:
> ```yaml
> safe-outputs:
> report-failure-as-issue: false
> ```
>
>

> Generated from [Prompt Optimization](https://github.com/githubnext/gh-aw-wizard/actions/runs/33684556510) · copilot · 36.7 AIC · [◷](https://github.com/search?q=repo%3Agithubnext%2Fgh-aw-wizard+is%3Aissue+%22gh-aw-workflow-id%3A+prompt-optimization%22&type=issues)

Contributor guide

No contributing guide indexed for this repository

Research direction

Start with .github/workflows/prompt-optimization.md and the failed run 33684556510, then read .github/skills/agentic-workflows/SKILL.md and the referenced gh-aw debug.md. Trace why the required Ollama models cannot be reached or loaded in the workflow. Done means Prompt Optimization can perform evaluation instead of reporting an incomplete result.

Written by the indexing model from the issue text.

Assessment

Tech stack
github-actions, huggingface, ollama
Domain
ai, ci-cd, devops
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Active
Clarity
Needs clarification
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.