shards-foundation/shards-agent-core
Curated, security-aware agent governance and repeatable local engineering workflows.
Use when checking that the local Shards agent harness can delegate read-only tasks with independent agent identities and verifiable receipts.
Use to validate and write a sanitized structured task checkpoint when transferring work between agents or sessions.
Use when planning or running repeatable maintenance of Shards Skills using approved workflow profiles and JSON evidence reports.
Use to check and explicitly repair Shards Skills catalogue metadata drift while preserving skill directory identity and provenance.
Use when separating orchestration control events from sub-agent result artifacts during bounded local skill execution.
Use when concurrently running three vetted read-only Shards checks through a bounded local sub-agent swarm.
Use to decide whether a local Python repository has enough verified evidence for release without publishing or deploying.
Use when inspecting the state, metadata and basic integrity of a local code repository before making changes.
Use before executing, importing or publishing an agent skill to inspect frontmatter, links, scripts, provenance and trust status.
Use when an agent must select and invoke a reviewed Shards skill through a bounded, auditable local dispatcher.
Use when delegating a bounded pair of independent read-only repository checks to uniquely identified local sub-agents.