Scope change: AI investigation is the product, drop zero-dependency constraint #64

New issue

Closed

opened 2026-04-11 09:37:03 -06:00 by claude-code · 0 comments

claude-code commented

2026-04-11 09:37:03 -06:00

Collaborator

Scope change

Two original design constraints are being dropped:

Zero-dependency Python CLI is no longer a goal. The tool installs from requirements.txt (or equivalent) like a normal Python project.
AI investigation is the headline. The base scan (tree, filetypes, code, recency, disk, report) becomes an internal first pass that feeds the agentic investigation. There is no --ai flag and no --no-ai mode. AI runs unconditionally.

Watch mode is being deleted as part of the same change. It exists to monitor filesystem churn without AI, which conflicts with the new philosophy. If a live-update mode is wanted later, it gets rebuilt as incremental AI re-investigation.

What changes

Code

Delete luminos_lib/watch.py.
Delete luminos_lib/capabilities.py and tests/test_capabilities.py. Move clear_cache() into cache.py.
luminos.py: remove --watch, --ai, --install-extras flags. AI runs unconditionally after the base scan. Keep --clear-cache, --fresh, -x, -d, -a, -o, --json. If ANTHROPIC_API_KEY is not set, exit 0 with a one-line hint instead of crashing.
luminos_lib/ai.py: drop the check_ai_dependencies() call and the import.
New requirements.txt listing anthropic, tree-sitter + grammars, python-magic.
Update setup_env.sh to install from requirements.txt.

Docs

Rewrite README.md to lead with AI investigation, drop the "two modes" framing and the watch feature line.
Update CLAUDE.md (project): rewrite Key Constraints, update module map.
Update PLAN.md: strip zero-dep philosophy notes from file map and phase descriptions that referenced lazy imports.
Update wiki (Architecture, DevelopmentGuide, Home).

Issue cleanup

Close #35 (incremental AI re-investigation in watch mode) as obsolete.

Out of scope for this issue

The AI investigation pipeline itself is unchanged. Phase 3 work (#19 through #29, #55, #56, #57) is unaffected.
Cost/budget controls, model selection (#34), and security mode (#32) are still future phases.
The MCP backend pivot (#39) is still the next architectural milestone after Phase 3.

Acceptance criteria

python3 luminos.py <target> runs the AI investigation end to end with no flag.
python3 luminos.py <target> with no ANTHROPIC_API_KEY exits cleanly with a one-line hint.
requirements.txt exists and setup_env.sh installs from it.
python3 -m unittest discover -s tests/ passes.
README, CLAUDE.md, PLAN.md, and the wiki reflect the new scope.
--ai, --no-ai, --watch, and --install-extras flags do not exist.
luminos_lib/watch.py and luminos_lib/capabilities.py are deleted.

## Scope change Two original design constraints are being dropped: 1. **Zero-dependency Python CLI** is no longer a goal. The tool installs from `requirements.txt` (or equivalent) like a normal Python project. 2. **AI investigation is the headline.** The base scan (tree, filetypes, code, recency, disk, report) becomes an internal first pass that feeds the agentic investigation. There is no `--ai` flag and no `--no-ai` mode. AI runs unconditionally. Watch mode is being deleted as part of the same change. It exists to monitor filesystem churn without AI, which conflicts with the new philosophy. If a live-update mode is wanted later, it gets rebuilt as incremental AI re-investigation. ## What changes **Code** - Delete `luminos_lib/watch.py`. - Delete `luminos_lib/capabilities.py` and `tests/test_capabilities.py`. Move `clear_cache()` into `cache.py`. - `luminos.py`: remove `--watch`, `--ai`, `--install-extras` flags. AI runs unconditionally after the base scan. Keep `--clear-cache`, `--fresh`, `-x`, `-d`, `-a`, `-o`, `--json`. If `ANTHROPIC_API_KEY` is not set, exit 0 with a one-line hint instead of crashing. - `luminos_lib/ai.py`: drop the `check_ai_dependencies()` call and the import. - New `requirements.txt` listing anthropic, tree-sitter + grammars, python-magic. - Update `setup_env.sh` to install from `requirements.txt`. **Docs** - Rewrite `README.md` to lead with AI investigation, drop the "two modes" framing and the watch feature line. - Update `CLAUDE.md` (project): rewrite Key Constraints, update module map. - Update `PLAN.md`: strip zero-dep philosophy notes from file map and phase descriptions that referenced lazy imports. - Update wiki (`Architecture`, `DevelopmentGuide`, `Home`). **Issue cleanup** - Close #35 (incremental AI re-investigation in watch mode) as obsolete. ## Out of scope for this issue - The AI investigation pipeline itself is unchanged. Phase 3 work (#19 through #29, #55, #56, #57) is unaffected. - Cost/budget controls, model selection (#34), and security mode (#32) are still future phases. - The MCP backend pivot (#39) is still the next architectural milestone after Phase 3. ## Acceptance criteria - `python3 luminos.py <target>` runs the AI investigation end to end with no flag. - `python3 luminos.py <target>` with no `ANTHROPIC_API_KEY` exits cleanly with a one-line hint. - `requirements.txt` exists and `setup_env.sh` installs from it. - `python3 -m unittest discover -s tests/` passes. - README, CLAUDE.md, PLAN.md, and the wiki reflect the new scope. - `--ai`, `--no-ai`, `--watch`, and `--install-extras` flags do not exist. - `luminos_lib/watch.py` and `luminos_lib/capabilities.py` are deleted.