The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and follow instructions, rules, and guidance provided by external CLI tool outputs (openspec instructions archive and openspec instructions specs). This creates a vulnerability where malicious or malformed data returned by the tool could influence the agent's behavior.
Ingestion points: The skill reads context, operationGuidance, and rules from the JSON output of the openspec CLI in Step 1 and Step 4.
Boundary markers: Present. The instructions explicitly direct the agent to keep external guidance separate from built-in steps, report conflicts, and avoid verbatim copying of these inputs into specs.
Capability inventory: The agent has access to the Bash tool and performs file system operations including mkdir and mv in Step 5.
Sanitization: Limited. While the skill includes instructions to ignore conflicting guidance, it lacks formal validation or sanitization of the values provided by the external tool before they are used in commands.
[COMMAND_EXECUTION]: The skill performs shell-based file system operations (mkdir -p, mv) using paths parsed dynamically from the openspec status CLI output (planningHome.changesDir, changeRoot). While this is the intended functionality, it relies on the integrity of the data provided by the CLI tool.

openspec-archive-change