sips-context-distiller

SIPS — Self-Improvement Plugin System

A self-improvement harness for Claude Code and ChatGPT / Codex. SIPS wraps a coding-agent session with memory-aware startup, safer tool use, verification hooks, bounded fresh-context delegation, and session closeout — so long-running sessions remember prior work, avoid repeated mistakes, recover from drift, and capture lessons before the context window disappears.

It does not try to make an agent smarter by swapping models. It improves the work loop around whatever model you already run.

SIPS 0.11 — visual evidence and explainable memory reuse. Compose tools with declared units, roles and context identities; inspect dependency drift; run bounded diagnostic policies against recorded probe outcomes. Frozen behavioral cases independently evaluate proposed policies. Passing tests never activate a change automatically. Workflow and migration guide.

Historical companion: Swift Harness, an experimental macOS control surface for the retired NCode setup. Its repository is marked RIP; compatibility with current SIPS is not established.

Each session can answer:

• What did we learn last time?
• Is this edit risky?
• Did the changed script still pass a smoke check?
• Is this task stuck enough to split out?
• What should future sessions remember?
• Which agent patterns have worked or failed before?

Suggested GitHub topics: agent-harness, ai-agents, claude-code, codex, memory-fabric, self-improvement, automation, evals, python, developer-tools

Start here

• Install the plugin in your host, then try the first-session checks
• Verify a source checkout before changing the harness
• Troubleshooting covers missing tools, stale installs, hooks, and state paths
• Architecture and Graph runtime docs explain the internals
• In-run improvement and visual evidence describe newer workflows
• Contributing, evaluation evidence, and decision history provide contributor context

Supported hosts

SIPS installs as a plugin in either harness, from the same repo:

• Claude Code — reads .claude-plugin/marketplace.json (marketplace sips-local).
• ChatGPT / Codex — reads .agents/plugins/marketplace.json (marketplace harness-local).

Delegation agents declare model: inherit, so they run on whatever model the host session already uses. The plugin never depends on a specific model or on model swaps; the gain comes from context control, scoped recall, verification, and recorded outcomes.

Install

Claude Code

/plugin marketplace add RasputinKaiser/Self-Improvement-Plugin

Then install harness-self-improvement from the marketplace UI, or use the marketplace name declared in this repository:

/plugin install harness-self-improvement@sips-local

ChatGPT / Codex

Register the local checkout as a marketplace in a Codex host with plugin support. The repository ships .agents/plugins/marketplace.json with marketplace name harness-local; select that marketplace and install:

harness-self-improvement@harness-local

For a staged local distribution, install.sh delegates to scripts/local_release.py. From a source checkout, this creates a new staging directory and a release.json receipt without installing into a host:

bash install.sh --stage /tmp/sips-local-stage

Choose a directory that does not already exist. The separate --install action changes the Codex cache and configuration, backs them up under the stage's recovery/ directory, and probes a fresh MCP subprocess. Review local_release.py before using it. A successful subprocess probe does not prove an already-open host session has refreshed.

First-session checks

After installation, restart or reload the host if it has not discovered the plugin. Begin with inspection before starting an improvement loop:

• On Codex: “Use $sips-control-plane to show SIPS status and available routes”
• On Codex: “Use $sips-proof-scanner to inspect this repository's verification surfaces”
• On Claude Code: use /recall to inspect relevant prior lessons, then /verify when you want to run harness checks

The control-plane skill uses homebase_status, homebase_routes, and freshness checks when native sips-homebase tools are available. If tools are deferred, discover the exact capability before treating it as missing. See troubleshooting for the source/cache/session boundary. Commands such as /improve, /goal, and /selfloop can change durable state; choose them deliberately for the task you want to run.

Requirements

Runtime scripts resolve $SIPS_HOME first, otherwise ~/.codex/sips. Retired host directories are never a runtime fallback. The SIPS-owned Memory Fabric subsystem is vendored in-repo — no external memory plugin is required.

What it adds

Adaptive core

observe → diagnose → propose → build → evaluate → ready_for_review

The active task agent writes the candidate; SIPS owns evidence, frozen evaluation, durable state, and review. Activation and rollback require an explicit reviewed candidate digest. The homebase_adaptation_read and homebase_adaptation_write MCP tools share the same controller as the CLI. NCode is retired; Codex is primary and Claude Code compatibility is retained.

Core workflow

A typical session looks like this:

1. Session starts.
2. The harness validates itself.
3. Memory Fabric health is checked.
4. Prior patterns and lessons are surfaced.
5. The user submits a task.
6. Relevant prior lessons are ranked and injected.
7. Before risky edits, preflight checks run.
8. After tool use, smoke checks and escalation checks run.
9. Before compaction, a continuity packet is written.
10. After compaction, session learnings are recorded.
11. At stop, the task outcome is captured for future pattern analysis.

The intent is to make each session leave behind useful state instead of vanishing into a transcript.

Slash commands