The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted web content (HTML text, metadata, and SERP summaries) to generate personalized link placement suggestions and outreach emails. This represents a vulnerability surface for indirect prompt injection if an attacker-controlled page contains malicious instructions designed to influence the LLM's drafting behavior.
Ingestion points: Untrusted data enters the context via the Website Content Crawler and Search Results datasets, specifically the text and markdown fields in the _wcc.json file.
Boundary markers: The instructions in SKILL.md emphasize reading the context to find natural link placements but do not specify explicit delimiters for the external text during the interpolation of data into drafting prompts.
Capability inventory: The skill has the capability to write files, perform network requests to the Apify API, and execute local shell commands (Node.js and Python helper scripts).
Sanitization: The skill implements a robust 'Skip Pass' (Step 6) with 11 specific rules to filter out adversarial mentions, non-editorial content, and off-category articles, which reduces the likelihood of processing malicious or irrelevant data.
[COMMAND_EXECUTION]: The skill uses several Node.js and Python scripts to manage the prospecting workflow. These scripts process JSON data, interact with the Apify API, and generate Excel reports. All commands are local and serve the primary purpose of the skill without implementing unsafe dynamic execution on user-supplied input.
[EXTERNAL_DOWNLOADS]: The skill interacts with external services, specifically Apify (via api.apify.com) and Ahrefs (via MCP tools). These interactions are documented, target well-known platforms, and are essential for the skill's functionality. The skill provides clear guidance on managing authentication tokens securely through environment variables.

apify-link-prospecting-outreach