The Agent Skills Directory

[SAFE]: The skill implements a robust 'Knowledge Verification Chain' designed to prevent hallucinations and technical inaccuracies.
Evidence: The framework established in SKILL.md and references/design.md enforces a strict hierarchy for information gathering (Codebase, Project Docs, Context7 MCP, Web Search) and explicitly instructs the agent to flag uncertainty rather than fabricate information.
[COMMAND_EXECUTION]: The skill utilizes standard command-line utilities for codebase analysis and version control management.
Evidence: Documentation in references/code-analysis.md and references/implement.md provides patterns for using ripgrep (rg), ast-grep (sg), and git to map existing repositories and perform atomic commits.
[EXTERNAL_DOWNLOADS]: The documentation references an installation script provided by the author's official organization.
Evidence: README.md includes the command npx @tech-leads-club/agent-skills install -s tlc-spec-driven to set up the skill environment.
[SAFE]: The skill incorporates persistent memory features to maintain state across sessions without using suspicious persistence mechanisms.
Evidence: State management is handled through a documented .specs/project/STATE.md file within the user's project directory, as described in references/state-management.md, rather than system-level persistence like cron jobs or shell profile modifications.

tlc-spec-driven