ui: tier cards — black background, new tier names (TRIAGE/STRATEGY/OPERATIONS)

SystemTruth rebrand: bigger UI, new diagrams, theme-cohesive HF Space

lfs: track docs/blog/*.png through git-lfs for HF Space compat

finalization: blog + README + execution rewrite, drop 3B + openclaw shim

notebook: broaden Cell 10 SFT-only load except clause

notebook: fix GRPO prompts — apply chat template before passing to TRL

notebook: install openenv-core (and other repo deps) in Cell 0

notebook: switch Cell 0 to Unsloth's official uv pattern + harden Cells 10-12

training: precision rewrite to prevent SFT collapse + GRPO variance starvation

notebook: use Qwen2.5-3B-Instruct (has chat template) not the base model

notebook: align with Unsloth-recommended TRL 0.22.2

notebook: show pip install progress (drop -q flag)

notebook: rewrite for robustness and resumability

notebook: fix TRL >=1.0 API breakage

hackathon sprint: grader collapse + coliseum rename + training pipeline

data: 30 expert episodes + training notebook

honesty pass: README + execution.md + manifests now match the codebase

UI rewrite: visual-spec terminal + held-out eval streamer

UI Build Addendum: mount Gradio at /, preserve full FastAPI surface

24-hour sprint: all 3 tiers runnable + MCP dual-route + terminal-style Gradio UI

docs: expand README + execution.md into comprehensive references

Tier-escalating sre-gym v3.0 — compute → horizon → realism

GRPO notebook: trajectory snapshots, Drive checkpointing, comparison table

Replace estimated baseline numbers with eval_sweep-verified ones

Position vs OpenEnv-hackathon competition + add baseline grid

Auto-fetch seed dataset + Open-in-Colab badges

Wire Colab Secrets bridge into both training notebooks

Expand seed_combined.jsonl to 200 samples across 3 teachers

Document expanded training pipeline in README

Add Groq driver, GRPO notebook, and eval-sweep harness

Add Llama-3.3-70B teacher episodes + merge into seed_combined.jsonl

Add Fireworks driver for teacher-data collection

Document the 4-step empty-prompt filter in seed README

Rewire sanity_run.ipynb to SFT on the 39-sample Claude seed

Rewrite README as comprehensive hackathon landing page

Add vibe-coded SaaS scenarios + Claude-teacher seed dataset

Harden env + ship Claude skill, OpenClaw-RL shim, training pipeline

Rename project to sre-engineer-llm

Initial commit of Unified Incident Env v2: Honest SRE Simulator

Require token/model in UI start and add run summary charts

Update README with current links, UI, and latest model scores

Make simple UI run full scenario suite like terminal flow

Improve simple console auto-run and remove token persistence

Fix simple console reset done/reward logging

Add simple terminal-style UI as default app entry

Fix /step action validation and query_logs shorthand

Autofill required web-step fields from valid example

Fix web step payload compatibility for Space UI

Prepare competition-ready submission

Keep public task scores inside strict validator bounds