The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is instructed to ingest and follow instructions from external files which could be modified by an attacker.
Ingestion points: Global rules at ~/.decodie/rules/*.md and project-specific rules at .decodie/rules/*.md.
Boundary markers: Absent. The instructions explicitly direct the agent to 'Treat their content as additional instructions controlling documentation output'.
Capability inventory: The skill can write files to the .decodie/ directory and execute shell commands for hashing and state summarization.
Sanitization: No validation or filtering is performed on the content of the rule files before they are integrated into the agent's context.
[COMMAND_EXECUTION]: The skill uses local shell utilities for functional tasks.
Hashing: Instructs the agent to run shasum -a 256 via a shell pipeline to generate content-based anchors for code references.
State Management: Includes a bash script scripts/summarize-index.sh that uses jq, grep, sed, and awk to parse JSON data and provide context to the agent.

decodie-explain