Three tools with complete input schemas and descriptions. git_context and stack_context have clear, actionable descriptions (120-140 chars). All parameters are typed (boolean/string) with descriptions. However, output schemas are entirely undocumented, LLMs cannot plan downstream calls or extract structured data. No error handling guidance. Parameter descriptions lack format constraints (e.g., 'since' accepts any string; should specify ISO 8601). No enum constraints despite boolean flags that could be clearer. Tool names follow verb_noun pattern but lack specificity, 'git_context' and 'stack_context' are generic; 'get_git_branch_context' would be clearer. command_help is a utility tool with minimal value to agents.

• Document output schemas for all three tools. Example for git_context: 'Returns {branch: string, commit: string, pr?: {number: int, title: string, url: string}, diffs?: string}'. This enables LLMs to chain tools and extract data.
• Add format constraints to 'since' parameter: 'ISO 8601 date (YYYY-MM-DD) or relative format (e.g., "7 days ago", "2 weeks ago")'. Prevents invalid input.
• Add error handling guidance to descriptions. Example: 'If not in a git repository, returns error. If GitHub API fails, returns partial branch info without PR details. Call stack_context to verify project structure.'
• Rename tools for clarity: git_context → get_git_branch_context, stack_context → analyze_tech_stack. Verb_noun naming helps LLMs parse intent.
• Add enum constraints to boolean parameters or document their interaction. Example: 'pullRequest=false overrides description, reviews, comments, labels, checks. Set pullRequest=true to include PR metadata.'

Inferred effective spec: <=2025-11-25.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry