菜单
Read Zulip stream/topic history for project context, summarize decisions, and recover requirements or action items. Use when task context may depend on prior Zulip discussion rather than only repository files.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
| name | read-zulip |
| description | Read Zulip stream/topic history for project context, summarize decisions, and recover requirements or action items. Use when task context may depend on prior Zulip discussion rather than only repository files. |
Read Zulip history with shared credentials, then turn the result into project context for the current task.
Use scripts/read_zulip.py. It resolves credentials in this order:
ZULIPRC~/.zuliprc~/.config/zulip/zuliprcIf a collaborator keeps credentials elsewhere, pass --config /path/to/zuliprc
explicitly.
Never commit Zulip credentials into this repository.
For this repository, start with:
python .claude/skills/read-zulip/scripts/read_zulip.py \
--url "https://quantum-info.zulipchat.com/#channels/591576/VibeYoga-Hackathon-QEC/general" \
--limit 20
If general returns no relevant messages, widen to the whole stream:
python .claude/skills/read-zulip/scripts/read_zulip.py \
--channel 591576 \
--limit 20
If you still need the active thread referenced by repo docs, check:
python .claude/skills/read-zulip/scripts/read_zulip.py \
--channel 591576 \
--topic "channel events" \
--limit 20
--limit 20.Recent messages from a stream/topic:
python .claude/skills/read-zulip/scripts/read_zulip.py \
--channel "VibeYoga-Hackathon-QEC" \
--topic "channel events" \
--limit 20
Local substring filter over the fetched window:
python .claude/skills/read-zulip/scripts/read_zulip.py \
--channel 591576 \
--limit 100 \
--contains benchmark
Machine-readable output:
python .claude/skills/read-zulip/scripts/read_zulip.py \
--channel 591576 \
--limit 20 \
--format json
--format json only when another tool/script will consume the result.Use when preparing an AutoQEC demo walkthrough, advisor presentation, hackathon pitch, recorded narration, or evidence-backed explanation of why the demos matter and why their outputs are correct.
Read the artifacts of a completed AutoQEC run and propose improvements to the framework code itself — DSL gaps, weak baselines, miscalibrated safety thresholds, prompt drift, env limitations, orchestration friction. Advisory only; never edits framework files. Use after a run completes when the user asks "what should we change before the next run?" or feels the system is plateauing.
Run the AutoQEC research loop on a given env YAML. Orchestrates the autoqec-ideator / autoqec-coder / autoqec-analyst subagents via the Agent tool, invokes the Runner CLI for training + evaluation, and writes history.jsonl + pareto.json. Use when the user asks "run AutoQEC on <env>", "start a research round", or provides an EnvSpec YAML.
Inspect a stalled or failed run, identify root cause (bad hyperparameter / NaN pattern / OOM / env misconfig), and recommend a fix. Does NOT apply fixes autonomously.
Read an entire runs/<id>/log.md and assess research narrative coherence, identify stuck hypotheses, detect overfitting signs, and write a review markdown. Use after a full run completes or when a loop has been stuck for many rounds.
Audit a predecoder checkpoint against holdout seeds. Runs independent_eval (3 fair-baseline guards) and interprets borderline cases with LLM reasoning. Use when a round produces a promising Δ_LER and the user wants to confirm it is not a reward-hacking artifact.