The Agent Skills Directory

[PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill implements a memory recall system that processes untrusted data which is then injected into the agent's prompt context.
Ingestion points: The recall.sh script reads Markdown files from the .woostack/memory/ directory based on path matching.
Boundary markers: Notes are rendered using a simple Markdown header (### [name]) without explicit instructions for the agent to ignore embedded commands or instructions within the retrieved content.
Capability inventory: The skill and its associated scripts have the capability to write files (lib.sh), execute shell commands (SKILL.md, recall.sh), and interact with the git repository (resolve-base.sh, scope-match.sh).
Sanitization: No escaping or validation is performed on the content of the memory notes before they are interpolated into the output stream for the agent.
[COMMAND_EXECUTION]: The skill relies on multiple local bash scripts (scripts/build-index.sh, scripts/recall.sh, scripts/graph.sh, etc.) to perform its core functions. These scripts utilize standard Unix utilities like grep, sed, awk, and git to manage the .woostack workspace.
[EXTERNAL_DOWNLOADS]: The skill references a sibling dependency on woostack-doctor (located at ../woostack-doctor/scripts/doctor.sh). This script is executed during the initialization and repair procedures.

woostack-init