More screenshots
Works with the AI Agents you already use
Agent Session History
Search, export and report across Claude Code, Codex CLI, Cursor and OpenCode. Local, read-only, no upload.
$10
Agent Session History
Example session with this skill installed
Search my history for the session where I fixed the webhook retry logic and show me the transcript.
- Read your context and instructions
- Compiled the agent session history
I have searched your local session history and found a matching entry in Claude Code.
Session Search Results
| Agent | Session ID | Project | Date | Match Preview |
|---|---|---|---|---|
| Claude | fixture-sess-1 | payment-gateway | 2026-11-12 | "Fixed exponential backoff in webhook retry..." |
Transcript: claude:fixture-sess-1
User: The webhook retries are failing immediately instead of backing off.
Assistant: Checking services/webhook_manager.py. The retry logic uses a fixed 1-second delay. I will implement exponential backoff with a jitter.
Tool Call (write_file): Applied changes to services/webhook_manager.py.
Assistant: I have updated the logic. Retries will now occur at 2s, 4s, 8s intervals.
User: Looks good, let's deploy to staging.
Next steps
- Run
python3 scripts/session_history.py export claude:fixture-sess-1 --format mdto save this transcript. - Use
python3 scripts/session_history.py report --date 2026-11-12to see other changes made that day.
Connects securely to your tools. The creator never sees your data.
About this skill
Your coding agents already keep a record of everything you did. Claude Code writes JSONL transcripts, Codex CLI writes rollouts, Cursor keeps a local key-value store, and OpenCode has a SQLite database. None of them lets you search across all of it.
Agent Session History builds one local, read-only index over those four sources and gives your agent a small CLI to:
- Search everything — user messages, assistant replies, reasoning blocks and tool output, with a snippet and the exact session to open.
- Read a transcript — the full timeline, optionally with thinking blocks and tool calls, re-read from the original source so nothing is truncated.
- Export — Markdown or JSON for any session.
- Generate a work report — your own requests grouped by day, agent and project, which is what you actually need for standups and timesheets.
- See where the time went — session counts per project across all agents.
Requirements
- Python 3.9 or newer (standard library only, no pip install for the core)
- At least one supported agent installed locally
- Codex CLI: sessions are zstd-compressed. Reading them needs Python 3.14+ (built-in zstd) or the optional zstandard package. Without either, Codex is skipped and the other sources keep working.
- Cursor: best-effort. It reads Cursor's internal store, which Cursor can change between releases. If a composer cannot be parsed it is skipped.
Privacy and safety
- Source files are only ever read; nothing is uploaded.
- The only file written is a local index at ~/.local/share/session-history/index.db.
- No network requests, no telemetry, no API keys.
- Credentials and account tables are never read.
How to use
python3 scripts/session_history.py stats python3 scripts/session_history.py sessions --limit 20 python3 scripts/session_history.py search "webhook retry" python3 scripts/session_history.py show codex:demo-codex-upload --tools --reasoning python3 scripts/session_history.py export claude:fixture-sess-1 --format md --out session.md python3 scripts/session_history.py report --date 2026-11-15
On Windows use python instead of python3.
What it is not
- Not a cloud service and not a sync tool.
- Does not modify, delete or migrate sessions.
- Does not cover Gemini CLI, Aider or other agents yet.
Local history only. Zstd-compressed Codex files need Python 3.14+ or the zstandard package. Cursor support is best-effort. No support for Gemini CLI or Aider.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 1 day ago
Requires