The Agent Skills Directory

[SAFE]: The skill provides automated synthesis of project decision history, following a well-defined local workflow.
[COMMAND_EXECUTION]: The skill executes local Python scripts (scripts/decision_graph.py) to build and query the decision index. These scripts are bundled with the skill and operate only on local data.
[DATA_EXPOSURE]: The skill reads from the project's knowledge vault and configuration files (e.g., ~/.lode/config.yaml). This access is required for its core functionality and no sensitive system credentials (like SSH or AWS keys) are targeted.
[PROMPT_INJECTION]: Analysis identified a surface for indirect prompt injection as the skill processes untrusted JSON data from the knowledge vault to generate reports. However, the risk is mitigated by the structured nature of the processing and the lack of dangerous capabilities.
Ingestion points: JSON files in {vault}/raw/weeks/ and {vault}/raw/artifacts/.
Boundary markers: Not explicitly implemented in the synthesis prompts to isolate data from instructions.
Capability inventory: Restricted to local file-writing of Markdown and JSON indices within the vault path.
Sanitization: No content-based filtering or sanitization of input data prior to synthesis.

roadmap