AgentStack
Browse Sign in
Browse Why AgentStack Sell Docs
Sign in
SKILL verified MIT Self-run

Search Conversations

skill-lohomi15-claude-search-conversations-search-conversations · by Lohomi15

Search previous Claude Code conversations for a keyword or topic and return matching context. Use when the user asks "where did we talk about X", "find the chat about Y", "search past conversations", or wants to recall context from a prior session.

— No reviews yet
0 installs
25 views
0.0% view→install

Install

$ agentstack add skill-lohomi15-claude-search-conversations-search-conversations

✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • ✓ Prompt-injection patterns
  • ✓ Secret / credential exfiltration
  • ✓ Dangerous shell & filesystem operations
  • ✓ Untrusted network calls
  • ✓ Known-malicious package signatures

What it can access

  • ✓ Network access No
  • ✓ Filesystem access No
  • ✓ Shell / process execution No
  • ✓ Environment & secrets No
  • ✓ Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

View the full security report →

Verified badge

Passed review? Show it. Paste this badge into your README, it links to the public security report.

AgentStack Verified badge Links to your public security report.
[![AgentStack Verified](https://agentstack.voostack.com/badges/verified.svg)](https://agentstack.voostack.com/security/report/skill-lohomi15-claude-search-conversations-search-conversations)

Reliability & compatibility

✓ Security review passed
0 installs to date
— no reviews yet
○ 5mo ago

Declared compatibility

Claude CodeClaude Desktop

Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.

Preview Execution monitoring

We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.

How agent discovery & health will work →
Are you the author of Search Conversations? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

search-conversations

Finds mentions of a keyword or phrase across every Claude Code session stored in ~/.claude/projects/*/*.jsonl and returns matching snippets with surrounding context.

When to use

  • User asks about a prior conversation: "where did we discuss X", "which chat was it where we talked about Y", "find the session about Z".
  • User wants to recover context from a past session that isn't in memory.
  • User references something you don't recognize and suspects it was discussed before.

How to run

Pass the query as a single argument. Quote multi-word phrases.

python3 ~/.claude/skills/search-conversations/search.py "nagarhole"
python3 ~/.claude/skills/search-conversations/search.py "coorg trip" -c 2 -n 10

Flags:

  • -c N — include N turns of context before/after each hit (default 1).
  • -n N — cap number of hits (default 20).
  • --chars N — truncate each snippet line to N chars (default 400).
  • --since YYYY-MM-DD — only include matches on or after this date.
  • --until YYYY-MM-DD — only include matches on or before this date.

Use date flags when the user references timing ("last week", "yesterday", "in April"). Resolve the relative date to YYYY-MM-DD before passing it in.

Output is grouped per hit: timestamp, session id (first 8 chars), project dir, a context window of [user]/[assistant] turns, and the full jsonl path.

What to do with results

  1. Summarize the matching conversations for the user — dates, what was discussed, which session(s).
  2. If the user wants more detail, read the jsonl path directly with the Read tool or grep it for surrounding turns.
  3. If there are no matches, say so plainly; don't fabricate recall.
  4. Offer to resume the best match (see below). Users usually want to jump back in, not just read a snippet.

Resuming a matched conversation

/resume is a built-in Claude Code slash command — it can't be invoked via Skill or Bash. Only the user can type it. So the flow is confirm → print command → user runs it.

  1. After presenting matches, pick the most likely target (usually the newest hit, or the one whose snippet best matches the query). If there are multiple strong candidates, list them briefly and ask which.
  2. Confirm with the user in one line: "Want to resume this one?"
  3. On yes, print the full session UUID with the exact command to run:

> Run /resume to jump back in.

Guidelines:

  • Always print the full UUID, not the truncated 8-char prefix shown in search output. Pull it from the sessionId field or the jsonl filename.
  • Do NOT attempt to call /resume yourself via Skill or Bash — it will fail.

Notes

  • Search is case-insensitive and substring-based.
  • Results are sorted newest first.
  • The current active session is included — matches there are just "earlier in this chat".

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet, be the first.

Versions

  • v0.1.0 Imported from the upstream source.