Capture an AI session trace
Records the prompts, reasoning, skills, scripts, tool calls, and decisions that led to a completed result, organized as a day-by-day trajectory.
OpenExp
OpenExp is an open-source framework and Claude skill for saving completed AI workflows as reusable “exps.” It helps users and teams replay proven paths in Claude Code by installing packs locally in `~/.claude/skills/`.
What OpenExp does
OpenExp is an open-source framework and Claude skill for saving an AI-assisted experience after it reaches a result, then reusing that experience later. The site defines an “exp” as the path that led to a result: prompts, reasoning, skills, scripts, decisions, and the trajectory between them.
The product is designed for workflows where an agentic task has already been solved once and the useful parts of that run should be preserved. An exp can be dropped into `~/.claude/skills/` and later referenced by Claude Code, which then looks for a matching pack and follows the trajectory that produced the outcome.
Core capabilities
Four-file exp format
Stores each exp in a four-file package so the record is split between machine-readable facts, the anonymized timeline, a human README, and a Claude skill file.
Team and personal sharing
Lets authors share packs by placing them in `~/.claude/skills/` and reusing them inside Claude Code.
Situation-based replay
Matches live situations against installed exps by category tokens and arc shape, then quotes the relevant day from the matching trajectory.
Reuse the captured workflow
Returns the skills and scripts used at the relevant step, so the next run can follow a proven path instead of starting over.
Open-source local operation
Works with a local-first runtime and an open-source MIT license, with no telemetry, third-party server, or API key required.
Where OpenExp fits
Reusing a proven workflow
A team solves a repeatable workflow, such as sales follow-up or analytics setup, and wants to preserve the exact path that worked so it can be reused later.
Agent-assisted replay
An individual wants Claude Code to follow the same decision path on a similar task instead of restarting from a blank prompt each time.
Sharing an exp with a team
A pack author wants to share a completed exp with teammates by publishing it as a Claude skill that others can install locally.
Documenting the working path
A practitioner wants to capture the tools and scripts used at a key step so the next run can reuse the same assets and sequence.
Cataloging completed experiments
Someone comparing multiple completed runs wants to organize results by trajectory, grade, and outcome rather than by prompt alone.
Pros and Cons
Pros
- Captures concrete execution details, including prompts, reasoning, skills, scripts, tool calls, and decisions.
- Uses a structured pack format that separates facts, timeline, human-readable notes, and the Claude skill.
- Supports sharing and reuse through `~/.claude/skills/` and Claude Code.
- Emphasizes local-first operation with no telemetry, third-party server, or API key required.
- Published as open source under MIT, with authorship and citation preserved.
Cons
- The pricing page is unavailable, so current pricing and packaging are not documented in the provided source.
- The catalog and replay flow depend on having an installed exp that matches the current situation, so it is most useful after a relevant result has already been captured.
FAQ
What is OpenExp used for?
OpenExp is a framework and Claude skill for saving an experience after an AI-assisted task is completed, then reusing that experience later by dropping it into `~/.claude/skills/` and referencing it in Claude Code.
What does an exp contain?
The source describes a four-file pack format: `meta.yaml` for facts, `trajectory.anonymized.yaml` for the raw day-by-day timeline, `README.md` for humans, and `SKILL.md` for Claude.
How does replay work?
The use page says the skill is installed once, then Claude Code can match a live situation to an installed exp by shape and category tokens, quote the relevant day from the trajectory, and reuse the skills and scripts used at that step.
Is OpenExp local-first?
The site presents OpenExp as open source under MIT, with local runtime on the user's machine and no API key required.
What does OpenExp cost?
The pricing page at `/pricing` returns a page-not-found response, so no pricing details are available from the provided source.
Quick Facts
- Category
- Developer Tool
- Primary workflow
- Save an AI session as an exp, then replay it in Claude Code
- Format
- Four files: `meta.yaml`, `trajectory.anonymized.yaml`, `README.md`, `SKILL.md`
- Installation
- Drop packs into `~/.claude/skills/`
- License
- MIT
- Site
- openexp.ai
Alternativas a OpenExp
AakarDev AI
AakarDev AI helps teams manage AI provider access, project-level setups, logs, and analytics from one dashboard. It supports BYOK workflows and lists providers including OpenAI, Google Gemini, Anthropic, Groq, Mistral AI, and Perplexity AI.
BookAI.chat
BookAI te permite chatear con tus libros usando IA simplemente proporcionando el título y el autor.
Skills Janitor
Skills Janitor is a GitHub-hosted set of slash commands for auditing, tracking, and managing Claude Code and OpenAI Codex skills. It helps users find duplicates, broken links, and unused skills, then clean them up with self-contained commands.
Benchspan
Benchspan is an AI agent security platform that discovers agents, blocks prompt injection and data exfiltration in real time, and supports pre-launch red teaming. It is aimed at teams running agents in production and includes Python and TypeScript SDKs.
Lasso
Lasso is an ecommerce product data platform for enriching catalog records, processing supplier files, generating product content, and monitoring competitors. It combines a web app with a REST API, SDK, and MCP server for teams and developers.
Codex Plugins
Codex Plugins bundle reusable skills, app integrations, and MCP servers into workflows you can install in the Codex app or use from Codex CLI. They help extend Codex with connected-service tasks, reusable instructions, and shared team workflows.