tegwick 4a82650ba5 Record live proof-of-concept run via glas-harness's gateway

Real ext.bwrap sandbox, real OpenRouter call (qwen/qwen-2.5-72b-instruct),
real 2-turn tool-calling loop, real verified commit. Credential reused
the OPENROUTER_API_KEY already in the environment via credentials.py's
env-var short-circuit -- no OpenBao round trip needed this time.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

2026-07-26 13:46:24 +02:00

3.1 KiB

Raw Blame History

INTENT

Why rein-openweights exists, its boundary, and what it must never become. Established by glas-harness docs/adr/ADR-001-rein-harness-family.md.

Why it exists

Every current agent harness in the ecosystem (Claude Code, Grok CLI, Codex/GPT CLI) is a frontier-model-vendor harness: it only drives that vendor's own model. There is no way to run a current open-weight model through an equivalent agentic tool-use loop (plan → tool call → observe → repeat, with a bounded tool policy) without building one from scratch per occasion.

rein-openweights exists to be that harness: one agentic loop, model supplied via OpenRouter, usable anywhere a rein-aharness-equivalent session is wanted but a frontier-vendor dependency is not — cost-sensitive workloads, offline/degraded-network tolerance, or simply comparing open-weight model capability against frontier baselines on the same task.

Governing principle

It is a rein — a concrete harness backend under glas-harness's router, implementing the same harness contract as rein-aharness (session lifecycle, tool dispatch + policy, sandbox consumption via sand-boxer, State Hub reporting). It does not reimplement any of that framework machinery itself; glas-harness owns the contract, this repo owns the open-weight-model-specific agentic loop and tool execution.

What it must never become

Not a model router. Choosing which OpenRouter model to use for a given task, pricing, and fallback is a caller/config concern (--model), not logic this repo owns. llm-connect remains available as an optional dependency for future needs (structured-JSON side calls, diagnostics/replay) but is not load-bearing for the base agentic loop — see glas-harness ADR-002 and src/rein_openweights/openrouter_client.py.
Not a second harness framework. Session semantics, tool policy schema, sandbox consumption, and audit reporting are glas-harness's contract (GLAS-WP-0001-T01) — this repo implements it, not forks it.
Not a sandbox provisioner. Isolation comes from sand-boxer via glas-harness, the same as every other rein.
Not tenant-specific logic. Anything specific to one consumer belongs in that consumer's configuration, not hardcoded here.

Status

Minimal agentic loop implemented and unit-tested (26 tests, all mocked at the network/subprocess boundary): tool surface, OpenRouter client, credential acquisition, commit-verified success criterion, State Hub reporting. See workplans/REIN-OW-WP-0001-bootstrap.md.

Live-proven (2026-07-26): run for real through glas-harness's gateway (glas_harness.reins.rein_openweights.ReinOpenWeights) inside a real ext.bwrap sandbox, against the real OpenRouter API (qwen/qwen-2.5-72b-instruct) — a 2-turn tool-calling loop produced a real, verified git commit. Credential reused the OPENROUTER_API_KEY already present in the environment (the same one llm-connect resolves), via this repo's own credentials.py env-var short-circuit — no OpenBao round trip needed for this run.

The glas-harness adapter (glas_harness/reins/rein_openweights.py, mirroring reins/rein_aharness.py) now exists and is what performed this run.