ec2bedrock-cli

Agent and CLI to manage EC2 instances running self-hosted LLM inference: warm up an instance and serve an LLM, then hibernate it on idle to power ongoing inference workloads such as ADLC. Built on ec2-cli for the underlying EC2 machine lifecycle.

What you get

• An agent-first CLI cited from teken (afi-cli) — the runtime package has no third-party dependencies.
• A mesh identity — culture.yaml (suffix + backend) and the matching prompt file (CLAUDE.md for backend: claude).
• The canonical guildmaster skill kit (11 skills) under .claude/skills/, vendored cite-don't-import. See docs/skill-sources.md.
• A build + deploy baseline — pytest, lint, the agent-first rubric gate, and PyPI Trusted Publishing wired into GitHub Actions.

Quickstart

uv sync
uv run pytest -n auto                 # run the test suite
uv run ec2bedrock whoami  # identity from culture.yaml (command is `ec2bedrock`; dist is `ec2bedrock-cli`)
uv run ec2bedrock learn   # self-teaching prompt (add --json)
uv run teken cli doctor . --strict    # the agent-first rubric gate CI runs

Milestone 1 — cost estimator (offline)

All commands below are offline / dry-run (no AWS calls). compare and decide encode the recommendation in the exit code: ec2 = 0, token-provider = 20, mixed = 21. Note that 0 doubles as "recommend EC2" and generic success — a router that needs an unambiguous verdict should parse --json (the recommendation/decision field) rather than rely on exit code 0 alone; the non-zero codes (20/21) are unambiguous, and 1/2 remain user/environment errors.

ec2bedrock profile list
ec2bedrock profile show qwen27b-awq-l40s
ec2bedrock estimate --profile qwen27b-awq-l40s --hours 8 --days 22
ec2bedrock compare --profile qwen27b-awq-l40s --input-tokens 300000000 --output-tokens 60000000 --provider-input-price 0.58 --provider-output-price 3.01
ec2bedrock decide --workload meeting-mode --input-tokens 3000000 --output-tokens 500000 --sensitive --json

CLI

Every command supports --json. Results go to stdout, errors/diagnostics to stderr (never mixed). Exit codes: 0 success, 1 user error, 2 environment error, 3+ reserved.

Make it your own

1. Rename the package ec2bedrock/ and the ec2bedrock-cli CLI/dist name throughout pyproject.toml, the package, tests/, sonar-project.properties, and this README.md. The name is hard-coded in ~100 places, so list every occurrence first — see the git grep discovery command in CLAUDE.md, the authoritative rename procedure.
2. Edit culture.yaml with your suffix and backend.
3. Rewrite CLAUDE.md for your agent and run /init.
4. Re-vendor only the skills you need from guildmaster (see docs/skill-sources.md).

See CLAUDE.md for the full conventions (version-bump-every-PR, the cicd PR lane, deploy setup).

License

Apache 2.0 — see LICENSE.