Multica Project & Resource Manager
Create, inspect, and debug Multica projects and their durable resources so agents always have the right context.
The skill explicitly restricts allowed-tools to "Bash(multica *)", reflecting least privilege. Side effects are clearly stated: creating/updating/deleting projects and resources mutates durable workspace state and affects future tasks, with a specific requirement to ask before changing local_directory. Data-flow transparency is good: resource injection into task briefs and .multica/project/resources.json is explained. The skill only allows multica CLI commands, not arbitrary Bash, reducing risk. Deduction: no rollback mechanism, no mention of sensitive data handling or dependency security; user confirmation is not comprehensive, only explicitly required for local_directory changes.
The skill's instructions are detailed and self-consistent, with CLI examples and a debugging section. However, it has not been executed, so key-path reproduction is unverified. Error handling and failure feedback on abnormal input are not described in the skill document, potentially shifting debugging cost to users. Deduction: static review cannot exceed 10, and error-handling details are missing.
The skill targets a clear scenario: creating, inspecting, updating, or debugging Multica projects and resources. It identifies the audience and use cases with Chinese trigger examples, indicating Chinese language support. Semantic triggers are precise. Environment fit: no mention of mainland-China network accessibility; core function may depend on overseas services that could be unreachable. Deduction: boundaries are clear but environment-fit evidence is limited, especially reachability from mainland China.
The skill is well-structured with quick start, core model, CLI, and debugging sections. However, it lacks installation and dependency notes, versioning and changelog, and explicit maintenance responsibility. License metadata is NOASSERTION, and provenance is unverified. Deduction: governance and versioning are incomplete; troubleshooting guide is missing.
The skill provides detailed CLI commands and a debugging path, which likely accomplishes the task. But no representative output is provided, and the value claim is unverified. Deduction: static review cannot exceed 7, and there is no evidence of directly usable output.
The skill references a source-map file pointing to specific code locations, providing auditable primary material. However, it does not include test suites or CI workflows that verify the skill's key paths. Deduction: lack of third-party execution evidence, and the source map only covers code structure, not sufficient for independent reproduction.
- This skill has only been statically reviewed; no execution has been performed, so hidden runtime errors may exist.
- Some commands (e.g., project resource add) persist changes to workspace state that affect future tasks; authorization should be careful.
- The repository license is NOASSERTION and the publisher is unverified; ensure compliance before use.
- Core functionality may rely on overseas services (e.g., GitHub, multica.ai) that may be unreachable from mainland China, limiting usability there.
- No rollback mechanism is provided; incorrect resource updates may be difficult to revert, so caution is advised.
What does this skill do, and when should you use it?
This skill operates on Multica projects and their resources. Projects are durable context containers that bind resources like GitHub repos or local directories, which are injected into future agent task briefs and .multica/project/resources.json. It provides a complete set of CLI commands for creating, updating, inspecting projects, and managing project resources, as well as guidance on referencing projects in comments via mention links. It helps users ensure agents receive the correct project context and debug context misconfigurations.
Runs multica CLI commands to manage projects and project resources. Specifically: list, get, create, update projects, set/clear start and due dates, change project status; list, add (supporting github_repo and local_directory types), update, and remove project resources; and advises on how to reference a project in a comment using the mention://project/<id> link form. It reads JSON output from multica project list/get and project resource list for verification.
- A developer wants to permanently bind a GitHub repository to a project so all future agent tasks have the repo context
- A user needs to create a new project with start and due dates to plan a time-boxed effort
- An agent keeps failing to access the project's repository, and the developer needs to debug the project resources
- A user wants the project description to serve as common context for all tasks, automatically injected when a task is bound to the project
- A user needs to reference a project in a comment with a clickable link rather than plain text
What are this skill's strengths and limitations?
- Resources are durable and affect future tasks, ensuring agents always have the correct context
- Provides clear CLI command set covering common project lifecycle and resource management operations
- Supports two resource types: GitHub repos and local directories
- Project description can be injected into task briefs, useful for setting project-wide rules
- Depends on the Multica platform and CLI; cannot be used standalone
- Resource mutations have durable side effects on future tasks, requiring caution and potential unintended consequences
- No test suite or cross-platform validation mentioned
- For non-GitHub or local directory resources, only JSON escape hatch, no shortcut commands
How do you install this skill?
This skill is distributed as part of the Multica repository under server/internal/service/builtin_skills/. After installing the Multica platform (via Homebrew or install script), the skill is deployed with it. The specific skill folder should be placed in your agent's skills directory, but the source does not provide exact copying steps.
How do you use this skill?
Trigger in an environment with the multica CLI. Example prompt: "Bind this GitHub repo to the project", and the skill guides you through commands like multica project resource add. The skill mandates using --output json for parseable results.