← Back to Skills Marketplace
3386
Downloads
2
Stars
12
Active Installs
1
Versions
Install in OpenClaw
/install prompt-log
Description
Extract conversation transcripts from AI coding session logs (Clawdbot, Claude Code, Codex). Use when asked to export prompt history, session logs, or transcripts from .jsonl session files.
Usage Guidance
Do not run any scripts or commands referenced by this skill until you can inspect them. The SKILL.md refers to scripts/extract.sh, but the package contains no code — ask the publisher for the script source or a vetted install method. Confirm 'jq' (and on macOS 'gdate' or compatible 'date') are available before running any extraction. Be aware that the examples point to reading session files in your home directory; those logs often contain sensitive prompts or secrets, so process them only in a controlled environment and avoid uploading them to untrusted services. If you proceed, request the extract.sh source, review it for network calls or data exfiltration, or run it in an isolated container or VM.
Capability Analysis
Type: OpenClaw Skill
Name: prompt-log
Version: 0.1.0
The skill's stated purpose is to extract conversation transcripts from local AI session log files (`.jsonl`) and write them to local markdown files. The `SKILL.md` instructions are clear, align with this purpose, and do not contain any evidence of prompt injection attempts against the agent, data exfiltration, malicious execution patterns, or persistence mechanisms. The skill relies on a local script (`scripts/extract.sh`, not provided) and standard utilities (`jq`, `date`), with all operations confined to local file system access for its stated function.
Capability Assessment
Purpose & Capability
The name and description (extract transcripts from .jsonl session logs) match the instructions' stated goal. However, the SKILL.md repeatedly tells you to run a 'bundled script' at scripts/extract.sh, yet this skill package contains no code files or scripts. The skill metadata also omits the runtime dependency 'jq' declared in SKILL.md. These inconsistencies mean what the skill claims to provide is not actually present in the package.
Instruction Scope
The instructions are narrowly scoped to reading local .jsonl session files and writing a markdown transcript (expected for the stated purpose). But because the instructions tell the agent to execute a script that isn't included, it's unclear what that script would do. The examples reference reading files in user home dirs (e.g., ~/.codex/, ~/.claude/, ~/.clawdbot/), which is expected for a transcript tool but raises the usual sensitivity note: session logs often contain sensitive prompts/keys and should be inspected before being processed or uploaded.
Install Mechanism
This is an instruction-only skill with no install spec or code files, which is low risk in principle. The problem is the documentation claims a bundled script but none is present — so the package provides no executable to perform the stated task. No on-package install means nothing will be written to disk by the skill itself, but you should not execute or fetch external scripts unless you can review them first.
Credentials
The skill declares no required environment variables or credentials (reasonable). SKILL.md does, however, list runtime dependencies ('jq' and optionally 'gdate') that are not reflected in the declared requirements — a minor but meaningful mismatch. There are no demands for unrelated credentials or config paths.
Persistence & Privilege
No special persistence is requested (always=false). The skill does not request to modify other skills or system settings and does not claim to store credentials or enable itself automatically.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install prompt-log - After installation, invoke the skill by name or use
/prompt-log - Provide required inputs per the skill's parameter spec and get structured output
Version History
v0.1.0
Initial release of prompt-log
- Extracts conversation transcripts from AI coding session logs (.jsonl) for Clawdbot, Claude Code, and Codex.
- Supports optional time filters (`--after`, `--before`) and custom output path.
- Generates a markdown transcript by default in `.prompt-log/`.
- Requires `jq` and optionally `gdate` on macOS.
Metadata
Frequently Asked Questions
What is Prompt Log?
Extract conversation transcripts from AI coding session logs (Clawdbot, Claude Code, Codex). Use when asked to export prompt history, session logs, or transcripts from .jsonl session files. It is an AI Agent Skill for Claude Code / OpenClaw, with 3386 downloads so far.
How do I install Prompt Log?
Run "/install prompt-log" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Prompt Log free?
Yes, Prompt Log is completely free (open-source). You can download, install and use it at no cost.
Which platforms does Prompt Log support?
Prompt Log is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Prompt Log?
It is built and maintained by thesash (@thesash); the current version is v0.1.0.
More Skills