session-search

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Session Search

会话搜索

Answers "which of my past sessions was about X" and then pulls that session's context in.
用于回答“我哪次过往会话是关于X的”,并调取该会话的上下文。

Step 1: Check the graph exists and is fresh

步骤1:检查图谱是否存在且为最新版本

bash
obsidian-wiki sessions-query "<topic>" --json
If this exits 1 with "run sessions-build first", tell the user and offer
/session-brain
. If
graph.json
is more than ~7 days old, mention it and offer a rebuild — but do not silently rebuild, since that is a multi-second operation the user did not ask for.
bash
obsidian-wiki sessions-query "<topic>" --json
如果该命令返回状态码1并提示“先运行sessions-build”,请告知用户并提供
/session-brain
命令选项。如果
graph.json
已超过约7天,需告知用户并提供重建选项——但请勿静默重建,因为这是一项耗时数秒的操作,且用户并未主动要求。

Step 2: Rank

步骤2:排序

The scoring already combines four signals, so take the ordering as given rather than re-ranking:
  • similarity — TF-IDF cosine against the session's text
  • cluster lift — a session inside the best-matching topic scores higher even if its own words never matched. This is why a session that never said "telemetry" can still surface for it.
  • bookmark boost — a human already flagged this session as worth keeping
  • time decay — 90-day half-life, applied with a floor so an old exact match still outranks a fresh weak one
Useful filters:
--project NAME
,
--cluster N
,
--since DATE
,
--top N
.
评分已结合了四个信号,因此直接使用给定的排序结果即可,无需重新排序:
  • 相似度 — 基于会话文本的TF-IDF余弦相似度
  • 集群提升 — 即使会话自身文本未匹配关键词,只要它属于最匹配的主题集群,评分也会更高。这就是为什么从未提及“telemetry”的会话仍可能因该主题被检索出来。
  • 书签加权 — 已被人工标记为值得保留的会话
  • 时间衰减 — 半衰期为90天,设置最低分值,确保旧的精确匹配仍优先于新的弱匹配
实用筛选参数:
--project NAME
、
--cluster N
、
--since DATE
、
--top N
。

Step 3: Present

步骤3:展示结果

Show the top ~5 as a compact table — title, project, date, topic, and the
why
string, which already explains the match. Do not dump the raw JSON at the user.
Two things must be stated honestly rather than glossed over:
  • Entries with
    loadable: false
    are history-only: the transcript has been pruned from disk and only the prompts survive. They are listed in
    unloadable
    with a reason. Say the transcript is gone; do not imply it can be retrieved.
  • If everything relevant is unloadable, answer from the prompt text that is there and say that is all that remains.
将排名前5的结果以简洁表格形式展示——包含标题、项目、日期、主题以及
why
字段(该字段已解释匹配原因)。请勿向用户输出原始JSON数据。
有两件事必须如实说明,不得含糊:
  • 标记为
    loadable: false
    的条目属于仅历史记录:会话记录已从磁盘中删除,仅保留提示内容。这些条目会在
    unloadable
    列表中附带原因说明。需告知用户会话记录已不存在,不得暗示可恢复。
  • 如果所有相关条目均无法加载,则基于现有的提示文本进行回答,并说明这是仅存的内容。

Step 4: Load — hand off, do not reimplement

步骤4:加载——转交处理,勿重复实现

should_load
holds at most 3 session ids worth opening, already filtered to ones with transcripts.
Prefer an existing loader skill if the user has one installed:
  • claude-session-load
    — loads a Claude session by id
  • bookmark-load
    — use when the hit is bookmarked
⚠️ Those skills live in the user's personal skills directory (
~/.claude/skills/
), not in this repo. They may not exist. Do not assume them and do not error if they are absent — fall back to reading the transcript directly:
bash
undefined
should_load
变量最多包含3个值得打开的会话ID,且已过滤为带有会话记录的ID。
如果用户已安装现有的加载skill,优先使用:
  • claude-session-load
    — 通过ID加载Claude会话
  • bookmark-load
    — 当匹配结果已被书签标记时使用
⚠️ 这些skill存储在用户的个人技能目录(
~/.claude/skills/
)中,不在本仓库内。它们可能不存在。请勿假设其存在,若不存在也无需报错——直接回退到读取会话记录:
bash
undefined

the path is in the query result as
transcript

路径在查询结果的
transcript
字段中

obsidian-wiki sessions-show <session-id> --pretty

The transcript is JSONL and can be tens of MB. **Never read one whole.** Filter to the human turns
first:

```bash
grep -c '' <transcript>   # size check before anything else
python3 -c "
import json,sys
for line in open(sys.argv[1], errors='replace'):
    try: r = json.loads(line)
    except ValueError: continue
    if r.get('type')=='user' and not r.get('isSidechain') and not r.get('isMeta'):
        c = (r.get('message') or {}).get('content')
        if isinstance(c,str) and not c.lstrip().startswith('<'): print('>', c[:400])
" <transcript>
obsidian-wiki sessions-show <session-id> --pretty

会话记录为JSONL格式,大小可达数十MB。**绝不要读取完整文件**。先筛选出用户的对话内容:

```bash
grep -c '' <transcript>   # 先检查文件大小
python3 -c "
import json,sys
for line in open(sys.argv[1], errors='replace'):
    try: r = json.loads(line)
    except ValueError: continue
    if r.get('type')=='user' and not r.get('isSidechain') and not r.get('isMeta'):
        c = (r.get('message') or {}).get('content')
        if isinstance(c,str) and not c.lstrip().startswith('<'): print('>', c[:400])
" <transcript>

Step 5: Answer

步骤5:生成回答

Synthesise from what you loaded and cite the session ids you used. If the answer came from a history-only session's prompts rather than a real transcript, say so — the user needs to know how much of the context you actually have.
基于加载的内容生成回答,并注明所使用的会话ID。如果回答仅来自仅历史记录会话的提示内容而非完整会话记录,需如实告知用户——用户需要了解你实际掌握的上下文程度。

Related

相关功能

  • /session-brain
    — build or refresh the graph, survey and name topics
  • wiki-query
    — search compiled vault knowledge; this skill searches raw session history instead
  • wiki-agent
    — pull a topic out of another agent's history and distil it into the vault
  • /session-brain
    — 构建或刷新图谱,梳理并命名主题
  • wiki-query
    — 搜索已整理的vault知识;而本skill搜索的是原始会话历史
  • wiki-agent
    — 从其他Agent的历史记录中提取主题并提炼到vault中