EchoProfile

Turn your AI conversation history into a personal AI profile.

Features · Quick Start · Figure Pools · Roadmap · Contributing

English · 中文

🪞 Your AI conversations are becoming a second profile of you.

Your chats with Codex, Cursor, Claude Code, and other AI tools form a long-term trace of how you think, learn, decide, and collaborate.

EchoProfile generates a personal AI profile from your AI conversation history.

✨ Before any AI remembers you on your behalf, you should own your own memory first.

Generate a CoPA Profile from local AI conversation history, across session, project, or global scopes.

EchoProfile can generate a profile from different scopes of AI conversation history:

• Single session: understand your expression and judgment in one collaboration
• Single project: observe how you think across a project lifecycle
• Global history: extract more stable patterns from long-term AI usage

The profiling method is inspired by CoPA (Cognitive Personalization Assessment), a cognitive-factor framework for personalization. It looks beyond “what you talked about” and asks:

• How do you build trust?
• How do you anchor a problem in context?
• How do you maintain structure in your thinking?
• How do you manage cognitive load?
• How do you use metacognition?
• What kind of response resonates with you most?

Citation: Hang Su, Zequn Liu, Chen Hu, Xuesong Lu, Yingce Xia, Zhen Liu. CoPA: Benchmarking Personalized Question Answering with Data-Informed Cognitive Factors. arXiv:2604.14773, 2026. https://arxiv.org/abs/2604.14773

Abstract profiles are boring!

You can project your AI profile into different figure pools:

• Scientists: which kind of scientist does your thinking style resemble?
• Entrepreneurs / investors: which kind of operator does your decision style resemble?
• MBTI anime archetypes: a lighter way to understand your expressive temperament

You can also import custom figure pools, and this repository includes a skill for creating them.

Just for fun!

Thought Echoes maps you into a figure pool so you can find your own echoes.

EchoProfile should not limit Thought Echoes to built-in figure pools.

This repository includes skills/figure-pool-generator, a skill dedicated to creating EchoProfile-compatible figure pools. You can ask AI to create a new Figure Pool around a theme, such as:

• Scientists from a specific era
• Chinese and American internet entrepreneurs
• Investors and founders
• Writers, philosophers, and artists
• Any custom group of characters or personality archetypes

We also welcome community-contributed figure pools: schools of thought, industries, fictional worlds, character groups, and anything you personally love can become a new reference system for Thought Echoes.

The skill helps generate figure-pool data that follows EchoProfile’s schema, and can package it as an importable zip when needed:

python3 skills/figure-pool-generator/scripts/validate_figure_pool.py --input src/data/figure-pools/<pool-slug>.json
python3 skills/figure-pool-generator/scripts/pack_figure_pool_zip.py --input src/data/figure-pools/<pool-slug>.json --output zip/<pool-slug>.zip

Generated zip files are placed under the repository-level zip/ directory, for example:

zip/<pool-slug>.zip

In EchoProfile’s Figure Pools page, you can upload this zip pool directly. The app reads its pool.json and portraits/ assets, then makes it available as a new Thought Echoes reference system.

• CoPA Profile: generate a user profile from AI conversations
• Profile Snapshot: save historical profile snapshots
• Markdown / JSON export
• Thought Echoes: map profiles into figure pools
• Figure Pools: import, manage, switch, and edit figure pools
• Session Board: multi-session timeline view
• Session search, browsing, and message rendering
• Token, tool-call, error, and workflow analysis
• Tauri desktop app with a local-first runtime

Configure an OpenAI-compatible model locally for generating CoPA Profile and Thought Echoes.

• Profile comparison across time
• Stronger long-term pattern recognition
• Blind-spot and repeated-behavior hints
• More shareable figure pools
• Exportable local AI memory / user profile
• Support for more AI tools and history sources

Download packaged builds from GitHub Releases, or run from source for development.

For AI agents and terminal workflows, install the EchoProfile CLI plus bundled agent skills:

curl -fsSL https://raw.githubusercontent.com/3kyou1/EchoProfile/main/install-agent.sh | sh

This installs:

• echo-profile, the provider-agnostic terminal CLI
• EchoProfile agent skills into SKILLS_DEST, or by default the Codex-compatible $CODEX_HOME/skills / ~/.codex/skills directory

Restart agents that load skills at startup after installation.

To install only the CLI:

curl -fsSL https://raw.githubusercontent.com/3kyou1/EchoProfile/main/install-cli.sh | sh

Verify provider discovery:

echo-profile version
echo-profile list providers

The CLI itself is provider-agnostic. EchoProfile currently supports local history from Aider, Claude Code, Cline, Codex CLI, Cursor, Gemini CLI, and OpenCode.

pnpm install
pnpm tauri:dev

If you want to use EchoProfile in a browser, build and start the WebUI Server:

just serve-build-run

If you have already built it, start it directly:

just serve

For WebUI Server development:

just serve-dev

Docker runs the WebUI Server, not the desktop client. For first-time setup, copy the environment template:

cp .env.example .env

Edit .env and set at least:

ECHOPROFILE_TOKEN=your-secret-token

Start the WebUI container:

docker compose up -d --build

Open after startup:

http://127.0.0.1:3727/?token=your-secret-token

By default Docker mounts ~/.claude, ~/.codex, and ~/.local/share/opencode. If your remote Linux host uses different paths, set CLAUDE_HOME, CODEX_HOME, or OPENCODE_HOME in .env.

pnpm dev

Note: running Vite alone is mainly for frontend UI debugging. Features that depend on Tauri / WebUI APIs may not be available.

pnpm build
pnpm test
pnpm lint

The echo-profile binary exposes a JSON-only CLI for agent skills and automation:

echo-profile version
echo-profile list providers
echo-profile list sessions --current-project
echo-profile profile collect --scope project --current-project --budget-chars 30000

profile collect only gathers local user-message text as structured JSON. It does not call an LLM or generate a profile; agent skills can use the returned messages as profile-generation input.

EchoProfile is especially open to contributions in these areas:

• New AI history importers
• CoPA Profile prompt and structure improvements
• Thought Echoes matching logic
• Figure-pool datasets
• Profile visualization
• Privacy-preserving data processing
• Multilingual documentation and UI

中文版文档见 README.zh-CN.md。

EchoProfile is an independent open-source project released under the Apache License 2.0.

It was initially derived from Claude Code History Viewer; the original MIT copyright notice and license text are preserved in NOTICE.