The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external data from codebase exploration and user-provided problem statements to generate milestone plans. This data is interpolated into prompts for multiple sub-agents, including drafting agents and adversarial critics.
Ingestion points: Untrusted data enters the agent context via the 'Exploration Digest' and 'Problem Brief' in Phase 1 and Phase 3 of SKILL.md.
Boundary markers: The skill uses Markdown headers (e.g., '## Problem Brief') to delimit sections, but it lacks explicit instructions or delimiters to prevent the sub-agents from obeying instructions embedded within the ingested data.
Capability inventory: The skill has the ability to write files to the local filesystem, dispatch sub-agents, and define shell commands for verification.
Sanitization: There is no evidence of sanitization, escaping, or validation of the external content before it is interpolated into the prompts.
[COMMAND_EXECUTION]: The skill identifies and records shell commands within the 'Verification Strategy' section. These commands are intended to be executed by a downstream process to validate milestone completion. If the command generation is influenced by malicious input during the exploration phase, it could lead to the execution of unauthorized commands, although this is mitigated by a mandatory user approval step in Phase 5.

milestone-planning