Context & Orientation

Use these tools to orient an AI assistant within your Compose project — from a quick overview to fine-grained context budgeting and session reports.

---

cp_insights

Return a concise project overview: module graph, Compose targets, preview count, and recommended next steps.

What to ask your AI:

• “what does this project look like?”
• “learn the codebase”
• “give me an overview of the app”

Parameters

Name	Type	Required	Default	Description
project_path	string	Yes	—	Absolute or relative path to the root of the target project

Tips

• Always a good first call at the start of a session — the response tells you exactly which other tools to reach for next.
• The output includes a module graph summary and flags whether the project uses Compose Multiplatform, Android-only Compose, or both.

---

cp_get_context

Return token-budgeted context about the project, scoped to the most relevant layer.

What to ask your AI:

• “show the project structure”
• “what are the Compose patterns used here?”
• “give me the full context”

Parameters

Name	Type	Required	Default	Description
project_path	string	Yes	—	Path to the root of the target project
scope	enum	No	full	One of: structure, compose, previews, patterns, full
max_tokens	int	No	—	Hard cap on tokens returned — useful when context window is tight
focus_files	array[string]	No	—	Limit context to these specific source files

Tips

• Use scope: structure when you only want the module/package layout without source details.
• Use scope: previews right before calling cp_render so the AI already knows which preview names are available.
• focus_files pairs well with scope: compose when reviewing a single screen’s composable hierarchy.

---

cp_configure_context

Persist a context profile for the project so subsequent cp_get_context calls respect the budget automatically.

What to ask your AI:

• “set context to performance mode”
• “switch to minimal context”

Parameters

Name	Type	Required	Default	Description
project_path	string	Yes	—	Path to the target project
context_profile	enum	Yes	—	One of: performance, balanced, minimal
custom_token_budget	int	No	—	Override the profile’s default token ceiling

Tips

• performance returns the richest context (good for architecture exploration); minimal strips everything except file names and preview counts.
• Profile is stored in .composeproof/context-profile.json and is gitignored by default — each developer can set their own.

---

cp_generate_report

Produce a self-contained HTML report of the session — screenshots, diffs, context summary, and any verification results.

What to ask your AI:

• “generate a report for this session”
• “summarize what we did”
• “create a shareable report”

Parameters

Name	Type	Required	Default	Description
project_path	string	Yes	—	Path to the target project
output_path	string	No	.composeproof/reports/report-<timestamp>.html	Where to write the HTML file
include_screenshots	bool	No	true	Embed rendered preview PNGs in the report
include_context	bool	No	true	Include the project context summary section
report_reason	string	No	—	Free-text label shown as the report title (e.g. “Pre-release visual audit”)

Tips

• The HTML file is fully self-contained (base64 images, inline CSS) — safe to attach to a PR or Jira ticket.
• Set report_reason to keep a clear audit trail when reviewing multiple features in one session.