agent-observability

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Agent Observability (Self-Reporting Log)

Agent可观测性（自我报告日志）

Operating rules

操作规则

Treat user corrections as issues to log.
Log locally in the current repo at
```
docs/observed-coding-agent-issues.md
```
.
Create the report file if it doesn’t exist.
Mandatory routing checklist (do not skip): (1) log the correction, (2) read the relevant skill file, (3) respond to the main task. This skill never handles the main task directly.
After logging, always open the relevant skill file before continuing, even for small/obvious tasks.
If the user request includes a concrete edit/implementation (README/docs/config/code), open
```
skills/coding/SKILL.md
```
before any questions, reads, or edits.
If the user asks for a plan/spec, open
```
skills/planning/SKILL.md
```
before drafting the plan.
If the user asks to create/update a skill or references
```
skills/
```
/
```
SKILL.md
```
, open
```
skills/skill-creator/SKILL.md
```
before continuing.
On multi-turn requests, re-evaluate routing at the start of each turn; logging does not replace skill invocation.

When opening references, use full repo paths like

skills/agent-observability/references/...

(not

references/...

When a reference trigger clearly matches, open the referenced file before drafting output.
If the user provides a word/length limit, keep the log entry short and minimize extra reads.
Rate limit: at most one log entry per session unless the user explicitly asks to log more.
Batch multiple corrections into a single entry by updating the existing root cause and guardrail notes; do not add verbatim quotes.

将用户的修正内容视为需要记录的问题。
在当前仓库的
```
docs/observed-coding-agent-issues.md
```
路径下本地记录。
若报告文件不存在则创建该文件。
强制路由检查清单（不可跳过）： (1) 记录修正内容，(2) 阅读相关Skill文件，(3) 响应主任务。本Skill从不直接处理主任务。
记录完成后，务必先打开相关Skill文件再继续，即使是微小/明显的任务也不例外。
若用户请求包含具体的编辑/实现（README/文档/配置/代码），在进行任何提问、阅读或编辑操作之前，先打开
```
skills/coding/SKILL.md
```
。
若用户要求制定计划/规格，在起草计划之前，先打开
```
skills/planning/SKILL.md
```
。
若用户要求创建/更新Skill或提及
```
skills/
```
/
```
SKILL.md
```
，在继续操作之前，先打开
```
skills/skill-creator/SKILL.md
```
。
对于多轮请求，每一轮开始时重新评估路由；记录操作不能替代Skill调用。
打开参考文件时，使用完整的仓库路径，如
```
skills/agent-observability/references/...
```
（而非
```
references/...
```
）。
当明确匹配参考触发条件时，在起草输出内容之前先打开参考文件。
若用户提供了字数/长度限制，保持日志条目简短并尽量减少额外的阅读操作。
频率限制：除非用户明确要求记录更多，否则每个会话最多记录一条日志条目。
将多个修正内容合并到单个条目中，更新现有的根本原因和防护措施说明；不要添加逐字引用。

Detect → Record → Resume

检测→记录→恢复

1) Detect correction

1) 检测修正内容

Trigger on phrases like or similar to:

“don’t do that”, “stop doing X”, “always do Y”, “never do Z” Friction alone is not sufficient; require a correction or explicit request.

触发条件为类似以下的语句：

“别那么做”、“停止做X”、“一直做Y”、“永远不要做Z” 仅存在摩擦不足以触发；需要明确的修正或请求。

2) Record issue (append)

2) 记录问题（追加）

Append a short entry to

docs/observed-coding-agent-issues.md

. If a log has already been written this session, batch new corrections under the same entry unless the user explicitly asks for another log.

Entry format:

Date/Time:
Task/Context:
What went wrong (root cause):
Proposed guardrail:

在

docs/observed-coding-agent-issues.md

中追加一条简短条目。若本次会话已写入过日志，将新的修正内容合并到同一条目中，除非用户明确要求另建日志。

条目格式：

日期/时间：
任务/上下文：
问题所在（根本原因）：
建议的防护措施：

3) Resume main task

3) 恢复主任务

Re-evaluate which skill applies next (coding/skill-creator/planning) and open that SKILL.md before responding. Do not continue the task until the relevant skill is loaded.

重新评估下一步适用的Skill（coding/skill-creator/planning），并在响应之前打开对应的SKILL.md文件。在加载相关Skill之前不要继续任务。

Routing shortcuts (read immediately)

路由快捷方式（立即阅读）

README/docs/config/code edits → read
```
skills/coding/SKILL.md
```
Plan/spec request → read
```
skills/planning/SKILL.md
```

Skill creation/update or

skills/

SKILL.md

mention → read

skills/skill-creator/SKILL.md

If correction + task appear in the same prompt, still read the task skill before any tool calls.

README/文档/配置/代码编辑 → 阅读
```
skills/coding/SKILL.md
```
计划/规格请求 → 阅读
```
skills/planning/SKILL.md
```

Skill创建/更新或提及

skills/

SKILL.md

→ 阅读

skills/skill-creator/SKILL.md

若修正内容和任务出现在同一个提示中，仍需在进行任何工具调用之前先阅读任务对应的Skill。

If unable to write

若无法写入

Output a “Self-report draft” block in the response with the entry content.
State that it was not written and needs manual application.

在响应中输出“自我报告草稿”块，包含条目内容。
说明该条目未写入，需要手动应用。

Reference triggers (open when clearly relevant)

参考触发条件（明确相关时打开）

Policy change / PR summary requested ->

skills/agent-observability/references/pr-template.md

政策变更 / 需要PR摘要 ->

skills/agent-observability/references/pr-template.md

References

参考文件

skills/agent-observability/references/pr-template.md

- PR summary template for policy changes

skills/agent-observability/references/pr-template.md

- 政策变更的PR摘要模板