The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external websites and search results to perform company research and define Ideal Customer Profiles (ICPs).
Ingestion points: Phase 1 (Research) utilizes web search and user-provided URLs to investigate target companies, market signals, and competitors.
Boundary markers: There are no explicit instructions to use delimiters or ignore-content markers when processing the data retrieved from external sources.
Capability inventory: The skill possesses the ability to execute local Python scripts (tam_builder.py), write to a Supabase database, and interact with external lead-generation APIs like Apollo and Crustdata.
Sanitization: The risk is significantly mitigated by mandatory user validation steps. The agent is instructed to present its synthesis for user approval before defining the ICP, and the TAM builder process requires a multi-stage approval flow (preview, then sample, then final build) before any data is written to the database.
[COMMAND_EXECUTION]: The skill executes structured shell commands to build market models.
It calls python3 skills/capabilities/tam-builder/scripts/tam_builder.py with specific flags (--preview, --sample, --test) and a configuration file.
These executions are based on predefined templates within the skill's infrastructure and are triggered only after explicit user confirmation.

icp-identification