Self Improving Opencode

A single starting skill that helps an AI agent get better through use. It turns verified work, user corrections, failures, and repeatable workflows into new or improved Agent Skills instead of making the user repeat the same steering.

This is self-improving AI in a deliberately basic, practical sense. It does not retrain the model. It improves the context and procedures the model receives on future tasks.

Memory and skills

The loop has two durable outputs:

• Memory stores facts, decisions, preferences, corrections, and ongoing context.
• Skills store reusable procedures, decision rules, tool usage, and verification.

It pairs naturally with GBrain: GBrain provides the memory layer, while this starting skill grows and refines the procedural skill layer. GBrain is optional; agents without it can use their existing memory mechanism.

Install

The skill defines what to learn. OpenCode, Claude Code, and Codex also have a completion adapter that runs the skill once after a substantial session. Follow the section for your client to install both parts.

OpenCode

Install the skill globally for OpenCode:

npx skills add https://github.com/hueyexe/self-improving-ai \
  --skill self-improving --agent opencode --global --yes

Append this entry to the existing plugin array in ~/.config/opencode/opencode.json. Keep your other plugin entries:

[
  "self-improving-opencode@latest",
  { "minMessages": 8, "gbrain": false }
]

Fully quit and restart OpenCode. OpenCode installs the package from the npm registry at startup and runs one completion review after a session reaches 8 messages.

Claude Code

Install the skill globally for Claude Code:

npx skills add https://github.com/hueyexe/self-improving-ai \
  --skill self-improving --agent claude-code --global --yes

Then add the plugin marketplace and install the completion adapter inside Claude Code:

/plugin marketplace add hueyexe/self-improving-ai
/plugin install self-improving@self-improving-ai

Start a new session after installation. The adapter runs one completion review after the transcript reaches 12,000 bytes.

Codex

Install the skill globally for Codex:

npx skills add https://github.com/hueyexe/self-improving-ai \
  --skill self-improving --agent codex --global --yes

Then add the plugin marketplace and install the completion adapter:

codex plugin marketplace add hueyexe/self-improving-ai
codex plugin add self-improving@self-improving-ai

Open /hooks in Codex and trust the new hook, then start a new thread. The adapter runs one completion review after the transcript reaches 12,000 bytes.

Other agents

Other clients can use the skill without an automatic completion adapter. Install it for every client supported by the Skills CLI:

npx skills add https://github.com/hueyexe/self-improving-ai --all --global

Eve and PromptScript currently report failures because they do not support global skill installation. Other supported clients still install successfully.

Configure

Each adapter runs at most once per session and skips short sessions. OpenCode defaults to 8 messages. Claude Code and Codex default to 12,000 transcript bytes.

Optional GBrain context

GBrain is off by default. In OpenCode, set gbrain to true in the plugin entry. For Claude Code or Codex, set this environment variable before launching the client:

export SELF_IMPROVING_GBRAIN=1

This does not install or configure GBrain. Follow the GBrain repository for setup. The flag only tells the review to use GBrain, when available, for durable non-procedural memory while keeping procedures in Agent Skills.

Threshold overrides:

export SELF_IMPROVING_MIN_TRANSCRIPT_BYTES=12000 # Claude Code and Codex
export SELF_IMPROVING_MIN_MESSAGES=8             # OpenCode env fallback

Set a threshold to 0 to review every session. Lower thresholds increase cost and make low-value skill churn more likely.

Manage the skill

List the installed skill and its source:

npx skills list --global

Update it from GitHub:

npx skills update self-improving --global --yes

Remove it:

npx skills remove self-improving --global --yes

Skill conventions

• skills/self-improving is the only starting skill.
• Skills created by the loop use clear capability-based names.
• Generated skills carry independent provenance, maturity, category, and scope metadata so they remain easy to organize and export.
• New workflows start as candidate, become validated after focused testing or successful reuse, and become proven after repeated use.
• Repository-specific procedures remain with their repository rather than being generalized into this collection.

The repository is the collection boundary. Skill metadata describes where each skill came from, what kind of skill it is, how mature it is, and where it belongs without forcing those details into its public name.

metadata:
  provenance: self-improving
  maturity: candidate
  category: workflow
  scope: global

Each directory under skills/ follows the Agent Skills specification.