scan-codebase

name	scan-codebase
description	Run the deterministic codebase scanner (core/components/plays/codify/lib/scan.py) over one or more repository roots and produce a bounded scan-index.json under STM. This skill is the sole structured-input producer for every infer-*-from-code skill. No LLM is involved — it shells out to scan.py and records status. Used exclusively by the /codify play.
version	0.1.0
user-invocable	false
model	sonnet
allowed-tools	Read, Write, Grep, Glob
deprecated	true
deprecated_note	#434 ProductOS realignment — superseded by the command model; retained for Phase E reference, not installed

scan-codebase

Called by the /codify play before any inference agent dispatches. Produces scan-index.json in STM. The index is the only structured-input that downstream inference skills consume.

Purpose

Garura's /codify play is a brownfield bootstrap — it reverse-engineers a product's LTM from an existing codebase. Before any LLM inference runs, a deterministic scan extracts structured facts: manifest data, directory tree, git signals, framework idioms, frontend detection, ADR harvest, config surface. That scan is this skill's job. Agents must never navigate source code as their primary input — they reason over the scan-index this skill produces.

Input

Receive from the /codify play orchestrator via JSON contract.

• repo_roots (list[path], required) — one or more absolute repository root paths to scan. Multi-repo monorepos pass multiple entries.
• output_path (path, required) — {stm_base}/{issue}/evidence/codify/scan-index.json
• size_limit_mb (int, optional, default 10) — soft cap on scan-index.json size; scan writes a partial index with scan_status: budget_exhausted on breach.
• time_limit_seconds (int, optional, default 600) — soft cap on wall-clock duration.
• extra_ignore (list[str], optional) — additional ignore globs beyond the built-in list (node_modules, dist, .git, etc.).

Process

1. Validate inputs

• Confirm every entry in repo_roots is an absolute path to an existing directory. Missing or non-directory → structured failure with what_failed: invalid_repo_root and the offending path.
• Confirm output_path parent directory is writable. If parent does not exist, create it. Failure to create → structured failure with what_failed: stm_path_unwritable.
• Confirm Python 3.8+ is on PATH. Missing → structured failure with what_failed: python_missing.
• Confirm scan.py exists at core/components/plays/codify/lib/scan.py relative to repo root. Missing → structured failure with what_failed: scan_script_missing.

2. Invoke scan.py

Build the command line:

python3 core/components/plays/codify/lib/scan.py \
    --output <output_path> \
    --size-limit-mb <size_limit_mb> \
    --time-limit-s <time_limit_seconds> \
    --path <root_1> [--path <root_2> ...] \
    [--ignore <pattern> ...]