The Agent Skills Directory

[DATA_EXFILTRATION]: The skill's mandatory protocol directs the agent to scan absolute local paths at D:\Projects\* to 'EXTRACT and adapt' code. This behavior targets sensitive development environments without explicit user consent or oversight, risking the exposure of private intellectual property.
[EXTERNAL_DOWNLOADS]: The skill requires the installation of the reasoningbank-agentdb package via npm, which is an external dependency not associated with a recognized trusted organization.
[REMOTE_CODE_EXECUTION]: The README and SKILL instructions direct the execution of npx ts-node setup-reasoningbank.ts. Because the script setup-reasoningbank.ts is not provided within the skill's distribution, this involves running unvetted external code at runtime.
[INDIRECT_PROMPT_INJECTION]: The skill's core workflow involves ingesting and evaluating reasoning trajectories and decision contexts. These inputs lack sanitization or protective boundary markers, allowing potentially malicious instructions in the data to influence the agent's behavior.
Ingestion points: trackTrajectory and judgeVerdict methods in SKILL.md and PROCESS.md.
Boundary markers: None present; the skill treats all trajectory data as trusted for pattern learning.
Capability inventory: Read, Write, Edit, Task, TodoWrite, Glob, Grep (all declared in SKILL.md).
Sanitization: None present; the skill extracts and consolidates patterns directly from reasoning text.

reasoningbank-adaptive-learning-with-agentdb