agent-launch (Python)
Idiomatic Python port of @p-vbordei/agent-launch. Draft platform-native release announcements (HN, Reddit, X, Mastodon, LinkedIn) from your repo's CHANGELOG + README. v0.1: drafts only — you post. Output is textually identical to the TS reference when given the same inputs.
What's in the box
draft_one(platform, project, context, repo, anthropic)— produce a platform-tuned post draft- Platforms:
hn,reddit,x,mastodon,linkedin gather_context(cwd, version)— collect CHANGELOG section + README + recent commitsload_launch_config(path)— strict pydantic-validatedlaunch.yaml- CLI:
agent-launch draft <version> [--platforms hn,x] [--out drafts/] - Strict input validation — no surprises in output
Install
pip install agent-launch
# or
uv add agent-launch
Quickstart
CLI (run inside any repo with a CHANGELOG.md, README.md, and launch.yaml):
cp launch.yaml.example launch.yaml && $EDITOR launch.yaml
ANTHROPIC_API_KEY=... agent-launch draft 0.2.0
# → launches/v0.2.0/{hn,reddit-*,x,mastodon,linkedin}.md
Library (offline — uses a scripted fake Anthropic client):
python examples/quickstart.py
See examples/quickstart.py for the 40-line walkthrough.
How it relates
| Repo | Lang | Install |
|---|---|---|
agent-launch |
TypeScript (reference) | npm i -g @p-vbordei/agent-launch |
agent-launch-py |
Python (this repo) | pip install agent-launch |
agent-launch-rs |
Rust | cargo install agent-launch |
All three share the same launch.yaml schema, the same vendored prompt templates, and the same conformance clauses.
Conformance
uv sync --extra dev
uv run pytest -v
The TS reference defines five conformance clauses (see SPEC.md §6). All pass here:
- C1 — Determinism. Same inputs → same output (with
temperature=0and a scripted client, byte-identical). - C2 — Length caps. Per-platform caps enforced with up to 2 retries; over-cap drafts marked
capped: false. - C3 — Repo URL. Every draft body contains the GitHub URL.
- C4 — No secrets / declared fields only. Env-var values are never echoed;
DraftResultexposes only declared fields. - C5 — Strict
launch.yaml. Missing/extra/invalid fields fail fast with a clear error.
Plus security clauses S2 (no tools field in the Anthropic call — model has no shell access), S4 (no sockets opened during drafting with a scripted client), S5 (--out sandboxed under cwd).
Architecture
See docs/architecture.md.
Development
git clone https://github.com/p-vbordei/agent-launch-py
cd agent-launch-py
uv sync --extra dev
uv run pytest -v
License
Apache-2.0
1 maintainer
Metadata
1 maintainer
Release files for agent-launch 0.1.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | |
|---|---|---|---|---|
| agent_launch-0.1.0.tar.gz | 85.0 kB | With provenance | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| agent_launch-0.1.0-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 105.4 kB
Release files / agent_launch-0.1.0.tar.gz
| Download URL | agent_launch-0.1.0.tar.gz |
|---|---|
| Size | 85.0 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
9fe76db05003b80e096cc2036cb4d8ccc8ed768a0fed87b931f8c43bd11c3321
|
|
BLAKE2b-256 checksum How to use checksums |
599f613aff4eed58d31c12b1802be49a624b35536c78a3d56dabf96b2d9602c8
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.12
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on May 20, 2026.
Transparency logRelease files / agent_launch-0.1.0-py3-none-any.whl
| Download URL | agent_launch-0.1.0-py3-none-any.whl |
|---|---|
| Size | 20.4 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
03d4b285a39453cccef248c3b80238fd119a68281cf7407dc9b8e87027948b80
|
|
BLAKE2b-256 checksum How to use checksums |
b5015d6d72cf1449618897f7cb7443d5c92a9f1f070d11ad6f28b72d9a24f828
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.12
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on May 20, 2026.
Transparency logSecurity for agent-launch 0.1.0
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
Full provenance
All 2 files in this release have been uploaded with provenance metadata.
Attestation for agent_launch-0.1.0.tar.gz
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on May 20, 2026.
Transparency logA note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.
