The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is configured to use the Bash tool and process.run operations as specified in the host-contract.json. These capabilities are constrained to resolving project context and performing semantic workspace operations defined by the host environment.
[INDIRECT_PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it ingests and processes various project records to generate proposals.
Ingestion points: Data is read from files such as PROJECT.md, TASKS.csv, TIME.csv, and INVOICES.csv (referenced in host-contract.json).
Boundary markers: The skill employs explicit Markdown markers () and utilizes SHA-256 checksums to seal the terms of issued proposals.
Capability inventory: The agent has access to Bash, Write, and Edit tools to manage the proposal documents and registry.
Sanitization: Security is enforced through a mandatory 'Native Mutation Sequence' (Inspect, Prepare, Verify, Apply) that requires byte-level validation and fresh readbacks before any record is finalized.

proposal