Fireworks Skill Memory

Organization

yizhiyanhua-ai

fireworks-skill-memory

Persistent cross-session experience memory for Claude Code and Codex skills. TRIGGER when user asks about skill memory, experience distillation, cross-session learning, skill knowledge injection, Claude/Codex memory, session-to-session improvement, or wants to install/configure fireworks-skill-memory.

Overview

Publisher	yizhiyanhua-ai
Repository	`fireworks-skill-memory`
Skill name	`fireworks-skill-memory`
Stars	93
Forks	7
Bundled files	34
License	MIT
Links	Skill source Repository Homepage

Markdown instructions
A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.
Works with any LLM
AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.
34 bundled files
Scripts, templates, and references the model can read while it works. Files are read-only and never executed.
Open source
Published by yizhiyanhua-ai on GitHub. Read the source before you install it.

Installation

Install the Fireworks Skill Memory AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

Open the app and go to Plugins → Skills.
Choose "Install from GitHub".
Paste the skill folder URL below and confirm.
https://github.com/yizhiyanhua-ai/fireworks-skill-memory/tree/main
Enable the skill in any chat where you want it available.

Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills

git clone --depth 1 https://github.com/yizhiyanhua-ai/fireworks-skill-memory.git \
  .claude/skills/fireworks-skill-memory

Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Fireworks Skill Memory in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Fireworks Skill Memory on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Fireworks Skill Memory is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

Loaded only when it is needed
The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.
Switch models mid-chat
Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

fireworks-skill-memory

Persistent experience memory for Claude Code and Codex skills. Shared memory core, runtime-specific adapters, skill-scoped lessons.

What It Does

Every coding-agent session starts from zero. The same mistakes repeat — wrong API parameters, broken sequences, proxy pitfalls — because the runtime has no durable skill memory between sessions.

fireworks-skill-memory solves this by:

Injecting past experience when a skill is invoked
Distilling new lessons into skill-scoped knowledge files
Keeping runtime-specific adapters thin so Claude hooks and Codex explicit flows share one memory core

Installation

Quick Install (Recommended)

In Claude Code, say:

"Help me install fireworks-skill-memory from https://github.com/yizhiyanhua-ai/fireworks-skill-memory"

Or run the one-command installer:

bash
curl -fsSL https://raw.githubusercontent.com/yizhiyanhua-ai/fireworks-skill-memory/main/install.sh | bash

npx skills Install

bash
npx skills add yizhiyanhua-ai/fireworks-skill-memory -g

After installing via npx skills, run the installer to set up hooks:

bash
curl -fsSL https://raw.githubusercontent.com/yizhiyanhua-ai/fireworks-skill-memory/main/install.sh | bash

Codex Setup

Run:

bash
./install-codex.sh

Then use:

bash
python3 cli/skill_memory.py inject --skill <skill-name>
python3 cli/skill_memory.py checkpoint --skill <skill-name> --note "..."
python3 cli/skill_memory.py flush --skill <skill-name> --summary-file ./session-summary.md

How It Works

Claude Code installs 4 hooks that run automatically:

Hook	Trigger	Script	Purpose
`PreToolUse`	Before Skill call	`pre-skill-inject.py`	Inject full KNOWLEDGE.md before skill executes
`PostToolUse`	After Read SKILL.md	`inject-skill-knowledge.py`	Inject top-N entries by relevance + capture error seeds
`PostToolUse`	After any tool call	`error-seed-capture.py`	Capture error signals to session-scoped file
`Stop`	Session end (async)	`update-skills-knowledge.py`	Distill new lessons via Haiku, update KNOWLEDGE.md

Codex uses explicit runtime commands instead:

Command	Purpose
`inject`	Load top-ranked lessons for a skill before a task
`checkpoint`	Save raw notes into the skill directory
`flush`	Distill explicit lesson sections from summary/session inputs into `KNOWLEDGE.md`

Data Flow


Skill invoked → PreToolUse injects experience → Claude executes with context
                                                    ↓
Session ends → Stop hook reads transcript → Haiku distills 1-3 lessons
                                                    ↓
                              KNOWLEDGE.md updated → Ready for next session

text
Codex task starts → inject loads top lessons → Codex executes with context
                                                ↓
Checkpoint/summary captured → flush distills explicit lessons
                                                ↓
                      KNOWLEDGE.md updated → Ready for next session

Knowledge Storage

text
<memory-home>/skills/<skill-name>/KNOWLEDGE.md   ← Distilled per-skill experience
<memory-home>/skills/<skill-name>/CHECKPOINTS.md ← Raw runtime notes
<memory-home>/global/KNOWLEDGE.md                ← Global cross-skill principles

Claude legacy runtime also uses:
~/.claude/skills/<skill-name>/KNOWLEDGE.md
~/.claude/skill-usage-stats.json
~/.claude/skill-memory.log

Each entry is tagged with [YYYY-MM] timestamp and [HIT:N] usage counter. Low-frequency, old entries are evicted first.

Configuration (Optional)

All settings are optional, configured via environment variables:

Variable	Default	Description
`SKILLS_KNOWLEDGE_MODEL`	`claude-haiku-4-5`	Default model for the Claude CLI distiller backend
`SKILLS_DISTILLER_BACKEND`	`claude-cli`	Distiller backend selector: `claude-cli`, `openai`, or `null`
`SKILLS_DISTILLER_DEBUG`	unset	Enable backend debug logging to the default distiller log file
`SKILLS_DISTILLER_LOG`	unset	Explicit path for distiller debug logs
`SKILL_MAX`	`100`	Max entries per skill
`GLOBAL_MAX`	`100`	Max global entries
`MIN_TOOL_CALLS`	`5`	Skip sessions with fewer calls (likely summaries)
`SKILLS_INJECT_TOP`	`20`	Max entries injected per active invocation

Requirements

Python 3.9+
Claude Code CLI for the automatic Claude runtime
Claude Haiku access for the default Claude CLI distiller backend
CODEX_HOME for the Codex runtime setup helper
OPENAI_API_KEY / OPENAI_BASE_URL / OPENAI_MODEL when using the OpenAI distiller backend
SKILLS_DISTILLER_DEBUG or SKILLS_DISTILLER_LOG when debugging backend failures

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

cli/skill_memory.py5 KB
CODE_OF_CONDUCT.md1 KB
docs/architecture-v2.svg12 KB
docs/architecture.svg11 KB
docs/harness-pattern.svg8 KB
docs/logo.svg2 KB
examples/settings-hooks.json1 KB
examples/skill-knowledge/browser-use.md1 KB
examples/skill-knowledge/find-skills.md1 KB
examples/skill-knowledge/hookify.md1 KB
examples/skill-knowledge/skill-adoption-planner.md1 KB
examples/skill-knowledge/skill-knowledge-extractor.md1 KB
examples/skill-knowledge/skill-roi-calculator.md1 KB
examples/skill-knowledge/skills-updater.md1 KB
examples/skill-knowledge/superpowers.md1 KB
examples/skill-knowledge/voice.md1 KB
index.html24 KB
install-codex.sh2 KB
install.sh8 KB
memory_core/__init__.py1 KB
memory_core/distiller.py12 KB
memory_core/store.py12 KB
memory_core/transcript.py4 KB
README.zh-CN.md13 KB
robots.txt1 KB
runtimes/claude_runtime.py1 KB
runtimes/codex_runtime.py4 KB
scripts/error-seed-capture.py3 KB
scripts/inject-skill-knowledge.py7 KB
scripts/pre-skill-inject.py2 KB
scripts/update-skills-knowledge.py18 KB
SECURITY.md1 KB
site.webmanifest1 KB
sitemap.xml1 KB

Frequently asked questions

What does the Fireworks Skill Memory AI skill do?

Why use Fireworks Skill Memory on TypingMind?

Because you install it once and use it with any model. Fireworks Skill Memory is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Fireworks Skill Memory in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/yizhiyanhua-ai/fireworks-skill-memory/tree/main. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Fireworks Skill Memory?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Fireworks Skill Memory?

As many as you like. As long as a model supports skills, you can use Fireworks Skill Memory with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Fireworks Skill Memory AI skill free?

Yes. It is published on GitHub by yizhiyanhua-ai under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Discrawl

openclaw

OrganizationPopular

Discord archive: search, sync freshness, DMs, summaries, TUI, repo/release work.

Gitcrawl

GitHub archive: issue/PR search, sync freshness, duplicate clusters, current GitHub handoff, and Gitcrawl repo work.

Graincrawl

Granola archive: search, sync freshness, notes, transcripts, panels, SQL counts, and Graincrawl repo work.

Openclaw Changelog Update

openclaw

OrganizationPopular

Regenerate OpenClaw release changelog sections from git history before beta, stable, or extended-stable releases.

Overview

Installation

Install in TypingMind

Install in another agent

Use it in TypingMind

Works with any AI model

Skill instructions

fireworks-skill-memory

What It Does

Installation

Quick Install (Recommended)

npx skills Install

Codex Setup

How It Works

Data Flow

Knowledge Storage

Configuration (Optional)

Requirements

More Information

Bundled files

Frequently asked questions

Related AI Skills

Set up your own AI workspace now