Repository navigation
v0.3.5 — Spectrum enforcement, goal SLO, and enterprise coat depth
AURA Harness — a runtime coat for agent loops: audit, policy, and compliance export.
v0.3.5 packages everything merged since v0.3.4. The headline is operational conformance on the spine: spectrum-level egress bind, goal drift and schedule SLO observer presets, and audit-report findings for declared intent vs observed behavior. This release also ships verified operator identity, hardened session export invariants, a public integrations index with an Ollama body loop, and a full documentation sweep.
Experimental (alpha). API may change before 1.0.
Install
pip install aura-harness==0.3.5 # or simply pip instal aura-harnessOptional Skillware integration:
pip install "aura-harness[skillware]==0.3.5"From source:
git clone https://github.com/ARPAHLS/aura.git
cd aura
git checkout v0.3.5
pip install -e ".[dev]"Spectrum & operational conformance
- Spectrum enforcement (#27) —
spectrum.levelinjects egress rules at session open (low/mid/high/full);sequencer_requiredat full bind;aura agent set --spectrum-level;effective_spectrumonagent show - Goal drift observer (#46) —
preset: goal_driftemitsconformance.driftwhen tool intents diverge from profile goal / forbidden topics - Schedule SLO observer (#46) —
preset: schedule_sloemitsslo.missedwhen required work misses deadline + grace; audit findingsGOAL_DRIFT/SCHEDULE_SLO_MISS - Host stress simulation —
scripts/aura_host_stress_sim.py(16 coat/spectrum scenarios) - Coat flow report —
scripts/aura_coat_flow_report.pystructured session breakdown per spectrum level
Identity & session hardening
- Verified operator identity (#55) — optional identity adapters (manual, mock, OIDC, Auth0);
identity.bound;ids.operatoron spine trailers; export redaction;aura identity show - Session lifecycle invariants (#15) — strict closed-session errors; atomic summary + OTel export; frozen open snapshot; in-memory audit report when
export=False
Integrations & examples
- Integrations layout (#19) — top-level
integrations/README.md; Skillware mock/live entrypoints; LangGraph stub - Ollama body loop (#20) — stdlib HTTP loop under
integrations/ollama/ - Audit pipeline example (#23) — two sessions, compare, hash-chain verify, OTel assert, CLI follow-ups
- Examples 10–11 — tailored metrics snapshot; goal drift + schedule SLO demo
- Skillware 0.5.4 sync — optional extra
skillware>=0.5.4,<0.6
Docs & comparison
- Documentation sweep (#14) — INDEX three-tier entry; refreshed architecture, concepts, field-services
docs/comparison.mdrefresh (#39) — loose / tight / tailored coat framing; shipped scope through v0.3.5- 140 tests — incl. goal/SLO observers, spectrum enforcement, audit pipeline OTel, example smoke
Links
Full diff: v0.3.4...v0.3.5