Collection of Harbor agents for AI evaluation
Project description
Harbor Agent: Claude Code with Skills
A Harbor agent that extends Claude Code with custom skills support. This library allows you to run Claude Code evaluations with pre-configured skills loaded into the container.
Installation
uv add harbor-agent-skilled-claude
Or with pip:
pip install harbor-agent-skilled-claude
Usage
Basic Usage
from harbor_agent.skilled_claude import ClaudeCodeWithSkills
from pathlib import Path
# Create agent with skills directory
agent = ClaudeCodeWithSkills(
logs_dir=Path("./logs"),
skill_dir=Path("./skills"), # Directory containing skill folders
)
Skill Filtering
# Load all skills (default)
agent = ClaudeCodeWithSkills(
logs_dir=Path("./logs"),
skill_dir=Path("./skills"),
skills=None, # Loads all valid skills
)
# Load specific skills only
agent = ClaudeCodeWithSkills(
logs_dir=Path("./logs"),
skill_dir=Path("./skills"),
skills="skill-a,skill-b", # Only loads skill-a and skill-b
)
# Baseline mode (no skills)
agent = ClaudeCodeWithSkills(
logs_dir=Path("./logs"),
skill_dir=Path("./skills"),
skills="", # Empty string = load no skills
)
With Harbor CLI
Use this agent with Harbor's run command by specifying it as the agent:
harbor run --agent harbor_agent.skilled_claude:ClaudeCodeWithSkills --task ./my-task
Skill Directory Structure
Skills should be organized in a directory structure like:
skills/
├── my-skill/
│ ├── SKILL.md # Required - skill definition
│ └── references/ # Optional - additional resources
│ └── guide.md
├── another-skill/
│ └── SKILL.md
Each skill directory must contain a SKILL.md file to be recognized as a valid skill.
Development
This project uses uv for package management.
Setup
# Clone the repository
git clone <repo-url>
cd harbor-agents
# Install dependencies
uv sync --all-extras
Running Tests
# Unit tests
uv run pytest tests/unit -v
# Integration tests (requires Harbor + API keys)
uv run pytest tests/integration -v
# All tests
uv run pytest -v
Linting and Type Checking
# Linting
uv run ruff check src tests
# Type checking
uv run mypy src
Building
uv build
License
MIT
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file harbor_agents-0.1.0.tar.gz.
File metadata
- Download URL: harbor_agents-0.1.0.tar.gz
- Upload date:
- Size: 205.2 kB
- Tags: Source
- Uploaded using Trusted Publishing? Yes
- Uploaded via: twine/6.1.0 CPython/3.13.7
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
42068a62fbe942d459205a18653acd613dbce06d77af1cd2a4743fff5c928a09
|
|
| MD5 |
b8e5e6da6bd918281d4cf740eac3822a
|
|
| BLAKE2b-256 |
c5865915b4dc43912f5522eec8dcf3e5894b6fb0ac1fdcc59899845cb2a709ae
|
Provenance
The following attestation bundles were made for harbor_agents-0.1.0.tar.gz:
Publisher:
publish.yml on rotemtam/harbor-agents
-
Statement:
-
Statement type:
https://in-toto.io/Statement/v1 -
Predicate type:
https://docs.pypi.org/attestations/publish/v1 -
Subject name:
harbor_agents-0.1.0.tar.gz -
Subject digest:
42068a62fbe942d459205a18653acd613dbce06d77af1cd2a4743fff5c928a09 - Sigstore transparency entry: 904319945
- Sigstore integration time:
-
Permalink:
rotemtam/harbor-agents@722edf68cbfb9f13181db456a1928fd4218c0e2a -
Branch / Tag:
refs/tags/v0.1.0 - Owner: https://github.com/rotemtam
-
Access:
private
-
Token Issuer:
https://token.actions.githubusercontent.com -
Runner Environment:
github-hosted -
Publication workflow:
publish.yml@722edf68cbfb9f13181db456a1928fd4218c0e2a -
Trigger Event:
push
-
Statement type:
File details
Details for the file harbor_agents-0.1.0-py3-none-any.whl.
File metadata
- Download URL: harbor_agents-0.1.0-py3-none-any.whl
- Upload date:
- Size: 8.0 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? Yes
- Uploaded via: twine/6.1.0 CPython/3.13.7
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
d846ff2f8722e36212e93863ec1f88a79e128cd5a13db1fb37b93083101a4f87
|
|
| MD5 |
09b0bde21c0960dc1f7b77287e3bb267
|
|
| BLAKE2b-256 |
7c36d5a2a0af89008c0c9df114f9ef3be6142911a7ae031b6058dc0a2dc6fb32
|
Provenance
The following attestation bundles were made for harbor_agents-0.1.0-py3-none-any.whl:
Publisher:
publish.yml on rotemtam/harbor-agents
-
Statement:
-
Statement type:
https://in-toto.io/Statement/v1 -
Predicate type:
https://docs.pypi.org/attestations/publish/v1 -
Subject name:
harbor_agents-0.1.0-py3-none-any.whl -
Subject digest:
d846ff2f8722e36212e93863ec1f88a79e128cd5a13db1fb37b93083101a4f87 - Sigstore transparency entry: 904320008
- Sigstore integration time:
-
Permalink:
rotemtam/harbor-agents@722edf68cbfb9f13181db456a1928fd4218c0e2a -
Branch / Tag:
refs/tags/v0.1.0 - Owner: https://github.com/rotemtam
-
Access:
private
-
Token Issuer:
https://token.actions.githubusercontent.com -
Runner Environment:
github-hosted -
Publication workflow:
publish.yml@722edf68cbfb9f13181db456a1928fd4218c0e2a -
Trigger Event:
push
-
Statement type: