The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes contents from implementation repositories and governance artifacts (specs, contracts, ADRs) to generate progress reports and revised plans.
Ingestion points: The evidence pass reads data from sibling implementation repositories (provided by the user) and the project's internal hsdd/ directory containing specifications and decision records.
Boundary markers: The instructions do not define delimiters or warnings to ignore embedded instructions within the audited files.
Capability inventory: The agent has permissions to write management documents to the filesystem, perform Git operations (commit, push), and generate prompts for other skills.
Sanitization: No sanitization or validation of the ingested content is specified before it is used to calibrate project velocity or determine the next steps in the execution plan.
[COMMAND_EXECUTION]: The skill requires active modification of the repository state through Git commands.
Evidence: The skill is instructed to commit management documents and push them to the spec-repo main branch within a submodule.
Context: These operations are fundamental to the skill's purpose of maintaining project synchronization and are gated by instructions to review the state first.

hsdd-checkpoint