[aw] Prompt Optimization reported incomplete result
维护者通常 1 天内回复
还没有人认领这个 Issue。
评估
- 难度
- 4/5
- 预计耗时
- 3-5 天
- 新手友好度
- 42/100
- Issue 类型
- 缺陷
- 描述清晰度
- 基本清楚
- 活跃度
- 活跃
- 技术栈
- github-actions, javascript, ollama
调研方向
Start by reading .github/workflows/prompt-optimization.md and the failed run linked in the issue. Trace the steps that restore the Ollama cache, fetch the GGUF models, and start the server; check whether each step succeeds and makes the CLI available. The issue is resolved when the workflow provides the evaluation server and both models so the optimization script can produce measured scores.
由索引模型根据 Issue 内容生成。
描述
Workflow Failure
Workflow: Prompt Optimization
Branch: main
Run: https://github.com/githubnext/gh-aw-wizard/actions/runs/38176105592
[!WARNING]
Task Could Not Be Completed: The agent reported that the task could not be performed due to an infrastructure or tool failure.
Reasons:
-
Required Ollama eval infrastructure was not available, so the hill-climbing loop could not run.
The task stated the Ollama cache was restored, GGUF proxies for both browser models were downloaded from Hugging Face, and an OpenAI-compatible server was started on (127.0.0.1/redacted) None of this was true in this run:
-
curl to (127.0.0.1/redacted) and /api/tags returned "Connection refused" (no listener on port 11434;
ss/netstatshowed nothing bound there). -
No
ollamaprocess was running (ps auxshowed only the Copilot CLI harness process). -
The
ollamaCLI was not on PATH (which ollama-> not found), though the binary exists at /opt/hostedtoolcache/ollama/0.33.2/x64/ollama. -
No model weights were cached anywhere on disk:
~/.ollamadoes not exist,OLLAMA_MODELSenv var is unset, and a full filesystem search for*.gguf,qwen2.5*, andsmollm2*found nothing related to the required models (hf.co/bartowski/Qwen2.5-1.5B-Instruct-GGUF:Q4_K_M and hf.co/unsloth/SmolLM2-360M-Instruct-GGUF:Q4_K_M).
Pre-optimization guard was checked first and found no open PR titled "Prompt optimization:" (gh pr list --state open returned []), so no review-comment action was applicable either.
Without a running eval server and the two GGUF proxy models, scripts/prompt-optimizer.mjs --evaluate/--score cannot produce real measured scores, so per the task's explicit constraints I did not fabricate scores, did not install/download the ~1-2GB of model weights myself (out of scope for this task and not something the task asked me to set up), and made no changes to src/js/slm.js. Working tree is clean (git status --short is empty); an incidental metadata diff in .github/skills/optimize-scenario-prompt/SKILL.md was reverted with git checkout --.
Recommended fix for the workflow owner: verify the upstream step(s) that are supposed to restore the Ollama cache, download the GGUF model proxies, and start ollama serve actually execute successfully before this job runs, and that ollama is placed on PATH or its full path is used.
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/38176105592
Manually invoke the agent
Debug this workflow failure using your favorite Agent CLI and the agentic-workflows prompt.
- Start your agent
- Load the
agentic-workflowsskill from.github/skills/agentic-workflows/SKILL.mdor https://github.com/github/gh-aw/blob/main/.github/skills/agentic-workflows/SKILL.md - Type
debug the agentic workflow prompt-optimization failure in https://github.com/githubnext/gh-aw-wizard/actions/runs/38176105592
[!TIP]
Stop reporting this workflow as a failure
To stop a workflow from creating failure issues, set
report-failure-as-issue: falsein its frontmatter:safe-outputs: report-failure-as-issue: false
Generated from Prompt Optimization · copilot · 30.5 AIC · ◷
- 主要语言
- JavaScript
- 星标
- 6
- 派生
- 1
- 平均合并
- 1 天 13 小时
- 30 天内合并 PR
- 21
环境准备
这个项目没有提供开发容器、Dockerfile 或贡献指南，环境需要你自己搭建：先看它的 README，通用步骤见我们的新手贡献指南。
从这里开始
- 先读完整个 Issue，再读项目的贡献指南。
- 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
- Fork 仓库，在一个分支上完成修改。
- 提交 Pull Request，并在描述里引用这个 Issue 编号。
githubnext/gh-aw-wizard 的其他 Issue
-
deep-research
难度 1/5 1 小时以内 新手友好度 65/100
githubnext/gh-aw-wizard#314 ·
维护者通常 1 天内回复
-
难度 3/5 1-2 天 新手友好度 25/100
githubnext/gh-aw-wizard#336 ·
维护者通常 1 天内回复
-
agentic-workflows
难度 1/5 1 小时以内 新手友好度 15/100
githubnext/gh-aw-wizard#335 ·
维护者通常 1 天内回复
-
agentic-workflows
难度 4/5 1-2 天 新手友好度 12/100
githubnext/gh-aw-wizard#334 ·
维护者通常 1 天内回复
-
agentic-workflows
难度 4/5 3-5 天 新手友好度 42/100
githubnext/gh-aw-wizard#333 ·
维护者通常 1 天内回复
查看 githubnext/gh-aw-wizard 的全部 Issue
相似的 Issue
-
BUG
难度 2/5 1-3 小时 新手友好度 74/100
维护者通常 1 天内回复
-
type/bug
难度 1/5 1 小时以内 新手友好度 75/100
flarum/framework#5231 · 1 条评论 ·
维护者通常 1 天内回复
-
难度 2/5 1-3 小时 新手友好度 82/100
apache/rocketmq-dashboard#6896 ·
维护者通常 3 天内回复
-
bug good first issue
难度 2/5 1-3 小时 新手友好度 84/100
维护者通常 1 天内回复
-
[workshop-sync] workshop/side-quest-17-06-output-injection.md: stale add_comment.cjs line references未关闭documentation
难度 2/5 1-3 小时 新手友好度 78/100
githubnext/gh-aw-workshop#4679 ·
维护者通常 1 天内回复