[spec-extractor] Update package specifications for constants, cli (run 1)

[github-actions]

Package Specification Updates

This PR updates README.md specifications for the following packages, analyzed during spec-extractor run 1.

Package	Status	Changes
constants	Updated	Added Crush engine, new feature flags, model env vars, version/port constants
cli	Updated	Added secrets set and secrets bootstrap subcommand entries
console	Unchanged	README already comprehensive and accurate
agentdrain	Unchanged	README already comprehensive and accurate

What Changed

pkg/constants/README.md

• Engine Constants: Added CrushEngine ("crush") and updated AgenticEngines to include it
• Feature Flags: Added 3 new flags: AwfDiagnosticLogsFeatureFlag ("awf-diagnostic-logs"), ByokCopilotFeatureFlag ("byok-copilot"), MCPCLIFeatureFlag ("mcp-cli")
• Model Env Vars: Added EnvVarModelAgentCrush, EnvVarModelDetectionCrush, CrushCLIModelEnvVar
• Version Constants: Added DefaultCrushVersion
• Network Port Constants: Added CrushLLMGatewayPort (10004)

pkg/cli/README.md

• Added gh aw secrets set subcommand entry (Create or update a repository secret)
• Added gh aw secrets bootstrap subcommand entry (Validate and configure all required secrets)

Extraction Method

• Source code analysis of exported symbols, types, and constants
• Cross-referenced with existing README content to avoid overwriting accurate sections
• Verified all constants against pkg/constants/*.go source files

Round-Robin State

• Packages processed this run: agentdrain, cli, console, constants
• Next packages in rotation: envutil, fileutil, gitutil, logger
• Total packages: 20
• Coverage: All 20 packages already have specifications; this run refreshes 4

---

Auto-generated by Package Specification Extractor workflow

Generated by Package Specification Extractor · ● 4.9M · ◷

• expires on Apr 20, 2026, 10:45 AM UTC

[utafrali]

The constants additions look structurally correct and follow existing patterns well, but the CLI README has two problems worth fixing: the secrets bootstrap entry incorrectly references secret_set_command.go (almost certainly the wrong file), and both new rows break the table's constructor-name convention by using file paths instead.

[utafrali]

secrets bootstrap is attributed to (secret_set_command.go), which is almost certainly wrong. Bootstrap and set are distinct operations and would typically live in separate source files (e.g. secret_bootstrap_command.go). If both commands actually live in the same file, that is worth a separate note, but the auto-extractor should verify this rather than defaulting the bootstrap entry to the set command's file.

[utafrali]

The new rows break the table's established convention for the middle column. Every other entry uses a Go constructor name (e.g. NewSecretsCommand, NewProjectCommand). The new entries use parenthesized file names like (secret_set_command.go) instead. If the extractor cannot determine the constructor name it should use a consistent placeholder (e.g. N/A) or leave the cell blank, not switch to a different identifier format mid-table.

[utafrali]

The existing AgenticEngines slice was documented in alphabetical order by engine name: ["claude", "codex", "copilot", "gemini"]. Alphabetically, "crush" (c-r) sorts before "gemini" (g), so the correct order would be []string{"claude", "codex", "copilot", "crush", "gemini"}. If the actual Go source appends "crush" at the end rather than inserting it in sorted position, the documentation is accurate but the Go slice itself is inconsistent with the prior ordering convention. Either way, this is worth a manual check against the source.