fix(cli): guard legacy hermes-agent entrypoint

[ooiuuii]

What does this PR do?

Fixes the legacy hermes-agent console script so metadata/probe invocations are side-effect free.

Today pyproject.toml maps hermes-agent directly to run_agent:main. A generated console script calls that function with no parsed CLI args, so hermes-agent --version, hermes-agent --help, and even bare hermes-agent can fall into the default Python 3.13 demo prompt and start a live agent/provider request.

This PR adds a small argparse wrapper for the legacy entry point and points the packaged script at it. The wrapper handles --version, --help, and no-argument invocations without calling main(), while preserving explicit runs through --query, positional prompt text, and the existing model/toolset flags.

Related note: #13712 also touches adjacent legacy version handling, but it is a broad 5k+ line integration PR. This PR keeps the fix narrow and reviewable.

Related Issue

Fixes #54648

Type of Change

• Bug fix (non-breaking change that fixes an issue)
• New feature (non-breaking change that adds functionality)
• Security fix
• Documentation update
• Tests (adding or improving test coverage)
• Refactor (no behavior change)
• New skill (bundled or hub)

Changes Made

• run_agent.py: add legacy_cli_main() and a legacy argparse parser for hermes-agent.
• pyproject.toml: point hermes-agent at run_agent:legacy_cli_main.
• tests/run_agent/test_legacy_cli_entrypoint.py: cover --version, --help, no args, explicit query dispatch, and packaged script target.
• tests/hermes_cli/test_verify_console_scripts.py: update the fake pyproject fixture to match the real script target.

How to Test

Reproduced on Windows with the installed hermes-agent.exe --version, which started the default Python 3.13 demo agent path and attempted a provider request before failing.

Validated the fix with:

1. uv run --extra dev python -m pytest tests\run_agent\test_legacy_cli_entrypoint.py -q
   • 5 passed
2. uv run --extra dev python -m pytest tests\hermes_cli\test_verify_console_scripts.py -q
   • 6 passed
3. uv run --extra dev python -m pytest tests\test_hermes_bootstrap.py::TestEntryPointsImportBootstrap::test_entry_point_imports_bootstrap -q
   • 6 passed
4. uv run --extra dev python -m ruff check run_agent.py tests\run_agent\test_legacy_cli_entrypoint.py tests\hermes_cli\test_verify_console_scripts.py
   • passed; ruff still reports an existing warning for an unrelated invalid # noqa directive in run_agent.py
5. python -m py_compile run_agent.py
6. Smoke-tested the generated worktree console script:
   • .venv\Scripts\hermes-agent.exe --version
   • .venv\Scripts\hermes-agent.exe --help
   • .venv\Scripts\hermes-agent.exe

I first tried the canonical bash scripts/run_tests.sh tests/run_agent/test_legacy_cli_entrypoint.py -q, but this Windows worktree had no .venv/venv and the script's POSIX $HOME/.hermes/hermes-agent/venv fallback did not match the local Windows install location.

Checklist

Code

• I've read the Contributing Guide / repo guidance
• My commit messages follow Conventional Commits
• I searched for existing PRs to make sure this isn't a duplicate
• My PR contains only changes related to this fix/feature
• I've run pytest tests/ -q and all tests pass
• I've added tests for my changes
• I've tested on my platform: Windows 11 Pro 10.0.22631, Python 3.11.15

Documentation & Housekeeping

• I've updated relevant documentation (README, docs/, docstrings) - N/A
• I've updated cli-config.yaml.example if I added/changed config keys - N/A
• I've updated CONTRIBUTING.md or AGENTS.md if I changed architecture or workflows - N/A
• I've considered cross-platform impact (Windows, macOS) - console script behavior is platform-independent; reproduced on Windows
• I've updated tool descriptions/schemas if I changed tool behavior - N/A

Screenshots / Logs

No raw failing provider output is included because the failing path can print credential-bearing request metadata. The regression tests and console-script smoke cover the side-effect-free behavior directly.

[teknium1]

Thanks for the focused fix. The premise is verified on current main: pyproject.toml:309 still maps hermes-agent directly to run_agent:main, while run_agent.py:5980-6008 initializes an agent, supplies the default Python 3.13 query when query is None, and runs the conversation. The generated local console-script shim likewise calls main() without parsing or forwarding arguments.

The PR's separate legacy_cli_main() wrapper and script-target change address that path without changing main()'s programmatic API. The wrapper covers all current main() parameters (run_agent.py:5839-5851) and its tests cover the requested metadata/no-argument and query-dispatch paths.

Automated hermes-sweeper review.