* fix(config): route CRUD through planningDir to honor GSD_PROJECT PR #1484 added planningDir(cwd) and the GSD_PROJECT env var so a workspace can host multiple projects under .planning/{project}/. loadConfig() in core.cjs (line 256) was migrated at the time, but the four CRUD entry points in config.cjs and the planningPaths() helper in core.cjs were left resolving against planningRoot(cwd). The result was a silent split-brain in any multi-project workspace: - cmdConfigGet, setConfigValue, ensureConfigFile, cmdConfigNewProject all wrote to and read from .planning/config.json - loadConfig read from .planning/{GSD_PROJECT}/config.json So `gsd-tools config-get workflow.discuss_mode` returned "unset" even when the value was correctly stored in the project-routed file, because the reader and writer pointed at different paths. planningPaths() carried a comment that "Shared paths (project, config) always resolve to the root .planning/" which described the original intent, but loadConfig() already contradicted that intent for config.json. project and config now both resolve through planningDir() so the contract matches the only function that successfully read config.json in the multi-project case. Single-project users (no GSD_PROJECT set) are unaffected: planningRoot() and planningDir() return the same path when no project is configured. Verification: in a workspace with .planning/projectA/config.json and GSD_PROJECT=projectA, `gsd-tools config-get workflow.discuss_mode` now returns the value instead of "Error: Key not found". Backward compat verified by running the same command without GSD_PROJECT in a single-project layout. Affected sites: - get-shit-done/bin/lib/config.cjs cmdConfigNewProject (line 199) - get-shit-done/bin/lib/config.cjs ensureConfigFile (line 244) - get-shit-done/bin/lib/config.cjs setConfigValue (line 294) - get-shit-done/bin/lib/config.cjs cmdConfigGet (line 367) - get-shit-done/bin/lib/core.cjs planningPaths.config (line 706) - get-shit-done/bin/lib/core.cjs planningPaths.project (line 705) * fix(template): emit project-aware references in template fill plan The template fill plan body hardcoded `@.planning/PROJECT.md`, `@.planning/ROADMAP.md`, and `@.planning/STATE.md` references. In a multi-project workspace these resolve to nothing because the actual project, roadmap, and state files live under .planning/{GSD_PROJECT}/. `gsd-tools verify references` reports them as missing on every PLAN.md generated by template fill in any GSD_PROJECT-routed workspace. Fix: route the references through planningDir(cwd), normalize via the existing toPosixPath helper for cross-platform path consistency, and embed them as `@<relative-path>` matching the phase-relative reference pattern used elsewhere in the file. Single-project users (no GSD_PROJECT set) get exactly the same output as before because planningDir() falls back to .planning/ when no project is active. Affected site: get-shit-done/bin/lib/template.cjs cmdTemplateFill plan branch (lines 142-145, the @.planning/ refs in the Context section). * fix(verify): planningDir for cmdValidateHealth and regenerateState cmdValidateHealth resolved projectPath and configPath via planningRoot(cwd) while ROADMAP/STATE/phases/requirements went through planningDir(cwd). The inconsistency reported "missing PROJECT.md" and "missing config.json" in multi-project layouts even when the project-routed copies existed and the config CRUD writers (now also routed by the previous commit in this PR) were writing to them. regenerateState (the /gsd:health --repair STATE.md regeneration path) hardcoded `See: .planning/PROJECT.md` in the generated body, which fails the same reference check it just regenerated for in any GSD_PROJECT-routed workspace. Fix: route both sites through planningDir(cwd). For regenerateState, derive a POSIX-style relative reference from the resolved path so the reference matches verify references' resolution rules. Also dropped the planningRoot import from verify.cjs since it is no longer used after this change. Single-project users (no GSD_PROJECT set) get the same paths as before: planningDir() falls back to .planning/ when no project is configured. Affected sites: - get-shit-done/bin/lib/verify.cjs cmdValidateHealth (lines 536-541) - get-shit-done/bin/lib/verify.cjs regenerateState repair (line 865) - get-shit-done/bin/lib/verify.cjs core.cjs import (line 8, dropped unused planningRoot)
484 lines
16 KiB
JavaScript
484 lines
16 KiB
JavaScript
/**
|
|
* Config — Planning config CRUD operations
|
|
*/
|
|
|
|
const fs = require('fs');
|
|
const path = require('path');
|
|
const { output, error, planningDir, withPlanningLock, CONFIG_DEFAULTS, atomicWriteFileSync } = require('./core.cjs');
|
|
const {
|
|
VALID_PROFILES,
|
|
getAgentToModelMapForProfile,
|
|
formatAgentToModelMapAsTable,
|
|
} = require('./model-profiles.cjs');
|
|
|
|
const VALID_CONFIG_KEYS = new Set([
|
|
'mode', 'granularity', 'parallelization', 'commit_docs', 'model_profile',
|
|
'search_gitignored', 'brave_search', 'firecrawl', 'exa_search',
|
|
'workflow.research', 'workflow.plan_check', 'workflow.verifier',
|
|
'workflow.nyquist_validation', 'workflow.ai_integration_phase', 'workflow.ui_phase', 'workflow.ui_safety_gate',
|
|
'workflow.auto_advance', 'workflow.node_repair', 'workflow.node_repair_budget',
|
|
'workflow.text_mode',
|
|
'workflow.research_before_questions',
|
|
'workflow.discuss_mode',
|
|
'workflow.skip_discuss',
|
|
'workflow._auto_chain_active',
|
|
'workflow.use_worktrees',
|
|
'workflow.code_review',
|
|
'workflow.code_review_depth',
|
|
'git.branching_strategy', 'git.base_branch', 'git.phase_branch_template', 'git.milestone_branch_template', 'git.quick_branch_template',
|
|
'planning.commit_docs', 'planning.search_gitignored',
|
|
'workflow.subagent_timeout',
|
|
'hooks.context_warnings',
|
|
'features.thinking_partner',
|
|
'context',
|
|
'features.global_learnings',
|
|
'learnings.max_inject',
|
|
'project_code', 'phase_naming',
|
|
'manager.flags.discuss', 'manager.flags.plan', 'manager.flags.execute',
|
|
'response_language',
|
|
]);
|
|
|
|
/**
|
|
* Check whether a config key path is valid.
|
|
* Supports exact matches from VALID_CONFIG_KEYS plus dynamic patterns
|
|
* like `agent_skills.<agent-type>` where the sub-key is freeform.
|
|
*/
|
|
function isValidConfigKey(keyPath) {
|
|
if (VALID_CONFIG_KEYS.has(keyPath)) return true;
|
|
// Allow agent_skills.<agent-type> with any agent type string
|
|
if (/^agent_skills\.[a-zA-Z0-9_-]+$/.test(keyPath)) return true;
|
|
// Allow review.models.<cli-name> for per-CLI model selection in /gsd-review
|
|
if (/^review\.models\.[a-zA-Z0-9_-]+$/.test(keyPath)) return true;
|
|
// Allow features.<feature_name> — dynamic namespace for feature flags.
|
|
// Intentionally open-ended so new flags (e.g., features.global_learnings) work
|
|
// without updating VALID_CONFIG_KEYS each time.
|
|
if (/^features\.[a-zA-Z0-9_]+$/.test(keyPath)) return true;
|
|
return false;
|
|
}
|
|
|
|
const CONFIG_KEY_SUGGESTIONS = {
|
|
'workflow.nyquist_validation_enabled': 'workflow.nyquist_validation',
|
|
'agents.nyquist_validation_enabled': 'workflow.nyquist_validation',
|
|
'nyquist.validation_enabled': 'workflow.nyquist_validation',
|
|
'hooks.research_questions': 'workflow.research_before_questions',
|
|
'workflow.research_questions': 'workflow.research_before_questions',
|
|
'workflow.codereview': 'workflow.code_review',
|
|
'workflow.review': 'workflow.code_review',
|
|
'workflow.code_review_level': 'workflow.code_review_depth',
|
|
'workflow.review_depth': 'workflow.code_review_depth',
|
|
'review.model': 'review.models.<cli-name>',
|
|
};
|
|
|
|
function validateKnownConfigKeyPath(keyPath) {
|
|
const suggested = CONFIG_KEY_SUGGESTIONS[keyPath];
|
|
if (suggested) {
|
|
error(`Unknown config key: ${keyPath}. Did you mean ${suggested}?`);
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Build a fully-materialized config object for a new project.
|
|
*
|
|
* Merges (increasing priority):
|
|
* 1. Hardcoded defaults — every key that loadConfig() resolves, plus mode/granularity
|
|
* 2. User-level defaults from ~/.gsd/defaults.json (if present)
|
|
* 3. userChoices — the settings the user explicitly selected during /gsd-new-project
|
|
*
|
|
* Uses the canonical `git` namespace for branching keys (consistent with VALID_CONFIG_KEYS
|
|
* and the settings workflow). loadConfig() handles both flat and nested formats, so this
|
|
* is backward-compatible with existing projects that have flat keys.
|
|
*
|
|
* Returns a plain object — does NOT write any files.
|
|
*/
|
|
function buildNewProjectConfig(userChoices) {
|
|
const choices = userChoices || {};
|
|
const homedir = require('os').homedir();
|
|
|
|
// Detect API key availability
|
|
const braveKeyFile = path.join(homedir, '.gsd', 'brave_api_key');
|
|
const hasBraveSearch = !!(process.env.BRAVE_API_KEY || fs.existsSync(braveKeyFile));
|
|
const firecrawlKeyFile = path.join(homedir, '.gsd', 'firecrawl_api_key');
|
|
const hasFirecrawl = !!(process.env.FIRECRAWL_API_KEY || fs.existsSync(firecrawlKeyFile));
|
|
const exaKeyFile = path.join(homedir, '.gsd', 'exa_api_key');
|
|
const hasExaSearch = !!(process.env.EXA_API_KEY || fs.existsSync(exaKeyFile));
|
|
|
|
// Load user-level defaults from ~/.gsd/defaults.json if available
|
|
const globalDefaultsPath = path.join(homedir, '.gsd', 'defaults.json');
|
|
let userDefaults = {};
|
|
try {
|
|
if (fs.existsSync(globalDefaultsPath)) {
|
|
userDefaults = JSON.parse(fs.readFileSync(globalDefaultsPath, 'utf-8'));
|
|
// Migrate deprecated "depth" key to "granularity"
|
|
if ('depth' in userDefaults && !('granularity' in userDefaults)) {
|
|
const depthToGranularity = { quick: 'coarse', standard: 'standard', comprehensive: 'fine' };
|
|
userDefaults.granularity = depthToGranularity[userDefaults.depth] || userDefaults.depth;
|
|
delete userDefaults.depth;
|
|
try {
|
|
fs.writeFileSync(globalDefaultsPath, JSON.stringify(userDefaults, null, 2), 'utf-8');
|
|
} catch { /* intentionally empty */ }
|
|
}
|
|
}
|
|
} catch {
|
|
// Ignore malformed global defaults
|
|
}
|
|
|
|
const hardcoded = {
|
|
model_profile: CONFIG_DEFAULTS.model_profile,
|
|
commit_docs: CONFIG_DEFAULTS.commit_docs,
|
|
parallelization: CONFIG_DEFAULTS.parallelization,
|
|
search_gitignored: CONFIG_DEFAULTS.search_gitignored,
|
|
brave_search: hasBraveSearch,
|
|
firecrawl: hasFirecrawl,
|
|
exa_search: hasExaSearch,
|
|
git: {
|
|
branching_strategy: CONFIG_DEFAULTS.branching_strategy,
|
|
phase_branch_template: CONFIG_DEFAULTS.phase_branch_template,
|
|
milestone_branch_template: CONFIG_DEFAULTS.milestone_branch_template,
|
|
quick_branch_template: CONFIG_DEFAULTS.quick_branch_template,
|
|
},
|
|
workflow: {
|
|
research: true,
|
|
plan_check: true,
|
|
verifier: true,
|
|
nyquist_validation: true,
|
|
auto_advance: false,
|
|
node_repair: true,
|
|
node_repair_budget: 2,
|
|
ui_phase: true,
|
|
ui_safety_gate: true,
|
|
ai_integration_phase: true,
|
|
text_mode: false,
|
|
research_before_questions: false,
|
|
discuss_mode: 'discuss',
|
|
skip_discuss: false,
|
|
code_review: true,
|
|
code_review_depth: 'standard',
|
|
},
|
|
hooks: {
|
|
context_warnings: true,
|
|
},
|
|
project_code: null,
|
|
phase_naming: 'sequential',
|
|
agent_skills: {},
|
|
};
|
|
|
|
// Three-level deep merge: hardcoded <- userDefaults <- choices
|
|
return {
|
|
...hardcoded,
|
|
...userDefaults,
|
|
...choices,
|
|
git: {
|
|
...hardcoded.git,
|
|
...(userDefaults.git || {}),
|
|
...(choices.git || {}),
|
|
},
|
|
workflow: {
|
|
...hardcoded.workflow,
|
|
...(userDefaults.workflow || {}),
|
|
...(choices.workflow || {}),
|
|
},
|
|
hooks: {
|
|
...hardcoded.hooks,
|
|
...(userDefaults.hooks || {}),
|
|
...(choices.hooks || {}),
|
|
},
|
|
agent_skills: {
|
|
...hardcoded.agent_skills,
|
|
...(userDefaults.agent_skills || {}),
|
|
...(choices.agent_skills || {}),
|
|
},
|
|
};
|
|
}
|
|
|
|
/**
|
|
* Command: create a fully-materialized .planning/config.json for a new project.
|
|
*
|
|
* Accepts user-chosen settings as a JSON string (the keys the user explicitly
|
|
* configured during /gsd-new-project). All remaining keys are filled from
|
|
* hardcoded defaults and optional ~/.gsd/defaults.json.
|
|
*
|
|
* Idempotent: if config.json already exists, returns { created: false }.
|
|
*/
|
|
function cmdConfigNewProject(cwd, choicesJson, raw) {
|
|
const planningBase = planningDir(cwd);
|
|
const configPath = path.join(planningBase, 'config.json');
|
|
|
|
// Idempotent: don't overwrite existing config
|
|
if (fs.existsSync(configPath)) {
|
|
output({ created: false, reason: 'already_exists' }, raw, 'exists');
|
|
return;
|
|
}
|
|
|
|
// Parse user choices
|
|
let userChoices = {};
|
|
if (choicesJson && choicesJson.trim() !== '') {
|
|
try {
|
|
userChoices = JSON.parse(choicesJson);
|
|
} catch (err) {
|
|
error('Invalid JSON for config-new-project: ' + err.message);
|
|
}
|
|
}
|
|
|
|
// Ensure .planning directory exists
|
|
try {
|
|
if (!fs.existsSync(planningBase)) {
|
|
fs.mkdirSync(planningBase, { recursive: true });
|
|
}
|
|
} catch (err) {
|
|
error('Failed to create .planning directory: ' + err.message);
|
|
}
|
|
|
|
const config = buildNewProjectConfig(userChoices);
|
|
|
|
try {
|
|
atomicWriteFileSync(configPath, JSON.stringify(config, null, 2), 'utf-8');
|
|
output({ created: true, path: '.planning/config.json' }, raw, 'created');
|
|
} catch (err) {
|
|
error('Failed to write config.json: ' + err.message);
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Ensures the config file exists (creates it if needed).
|
|
*
|
|
* Does not call `output()`, so can be used as one step in a command without triggering `exit(0)` in
|
|
* the happy path. But note that `error()` will still `exit(1)` out of the process.
|
|
*/
|
|
function ensureConfigFile(cwd) {
|
|
const planningBase = planningDir(cwd);
|
|
const configPath = path.join(planningBase, 'config.json');
|
|
|
|
// Ensure .planning directory exists
|
|
try {
|
|
if (!fs.existsSync(planningBase)) {
|
|
fs.mkdirSync(planningBase, { recursive: true });
|
|
}
|
|
} catch (err) {
|
|
error('Failed to create .planning directory: ' + err.message);
|
|
}
|
|
|
|
// Check if config already exists
|
|
if (fs.existsSync(configPath)) {
|
|
return { created: false, reason: 'already_exists' };
|
|
}
|
|
|
|
const config = buildNewProjectConfig({});
|
|
|
|
try {
|
|
atomicWriteFileSync(configPath, JSON.stringify(config, null, 2), 'utf-8');
|
|
return { created: true, path: '.planning/config.json' };
|
|
} catch (err) {
|
|
error('Failed to create config.json: ' + err.message);
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Command to ensure the config file exists (creates it if needed).
|
|
*
|
|
* Note that this exits the process (via `output()`) even in the happy path; use
|
|
* `ensureConfigFile()` directly if you need to avoid this.
|
|
*/
|
|
function cmdConfigEnsureSection(cwd, raw) {
|
|
const ensureConfigFileResult = ensureConfigFile(cwd);
|
|
if (ensureConfigFileResult.created) {
|
|
output(ensureConfigFileResult, raw, 'created');
|
|
} else {
|
|
output(ensureConfigFileResult, raw, 'exists');
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Sets a value in the config file, allowing nested values via dot notation (e.g.,
|
|
* "workflow.research").
|
|
*
|
|
* Does not call `output()`, so can be used as one step in a command without triggering `exit(0)` in
|
|
* the happy path. But note that `error()` will still `exit(1)` out of the process.
|
|
*/
|
|
function setConfigValue(cwd, keyPath, parsedValue) {
|
|
const configPath = path.join(planningDir(cwd), 'config.json');
|
|
|
|
return withPlanningLock(cwd, () => {
|
|
// Load existing config or start with empty object
|
|
let config = {};
|
|
try {
|
|
if (fs.existsSync(configPath)) {
|
|
config = JSON.parse(fs.readFileSync(configPath, 'utf-8'));
|
|
}
|
|
} catch (err) {
|
|
error('Failed to read config.json: ' + err.message);
|
|
}
|
|
|
|
// Set nested value using dot notation (e.g., "workflow.research")
|
|
const keys = keyPath.split('.');
|
|
let current = config;
|
|
for (let i = 0; i < keys.length - 1; i++) {
|
|
const key = keys[i];
|
|
if (current[key] === undefined || typeof current[key] !== 'object') {
|
|
current[key] = {};
|
|
}
|
|
current = current[key];
|
|
}
|
|
const previousValue = current[keys[keys.length - 1]]; // Capture previous value before overwriting
|
|
current[keys[keys.length - 1]] = parsedValue;
|
|
|
|
// Write back
|
|
try {
|
|
atomicWriteFileSync(configPath, JSON.stringify(config, null, 2), 'utf-8');
|
|
return { updated: true, key: keyPath, value: parsedValue, previousValue };
|
|
} catch (err) {
|
|
error('Failed to write config.json: ' + err.message);
|
|
}
|
|
});
|
|
}
|
|
|
|
/**
|
|
* Command to set a value in the config file, allowing nested values via dot notation (e.g.,
|
|
* "workflow.research").
|
|
*
|
|
* Note that this exits the process (via `output()`) even in the happy path; use `setConfigValue()`
|
|
* directly if you need to avoid this.
|
|
*/
|
|
function cmdConfigSet(cwd, keyPath, value, raw) {
|
|
if (!keyPath) {
|
|
error('Usage: config-set <key.path> <value>');
|
|
}
|
|
|
|
validateKnownConfigKeyPath(keyPath);
|
|
|
|
if (!isValidConfigKey(keyPath)) {
|
|
error(`Unknown config key: "${keyPath}". Valid keys: ${[...VALID_CONFIG_KEYS].sort().join(', ')}, agent_skills.<agent-type>, features.<feature_name>`);
|
|
}
|
|
|
|
// Parse value (handle booleans, numbers, and JSON arrays/objects)
|
|
let parsedValue = value;
|
|
if (value === 'true') parsedValue = true;
|
|
else if (value === 'false') parsedValue = false;
|
|
else if (!isNaN(value) && value !== '') parsedValue = Number(value);
|
|
else if (typeof value === 'string' && (value.startsWith('[') || value.startsWith('{'))) {
|
|
try { parsedValue = JSON.parse(value); } catch { /* keep as string */ }
|
|
}
|
|
|
|
const VALID_CONTEXT_VALUES = ['dev', 'research', 'review'];
|
|
if (keyPath === 'context' && !VALID_CONTEXT_VALUES.includes(String(parsedValue))) {
|
|
error(`Invalid context value '${value}'. Valid values: ${VALID_CONTEXT_VALUES.join(', ')}`);
|
|
}
|
|
|
|
const setConfigValueResult = setConfigValue(cwd, keyPath, parsedValue);
|
|
output(setConfigValueResult, raw, `${keyPath}=${parsedValue}`);
|
|
}
|
|
|
|
function cmdConfigGet(cwd, keyPath, raw, defaultValue) {
|
|
const configPath = path.join(planningDir(cwd), 'config.json');
|
|
const hasDefault = defaultValue !== undefined;
|
|
|
|
if (!keyPath) {
|
|
error('Usage: config-get <key.path> [--default <value>]');
|
|
}
|
|
|
|
let config = {};
|
|
try {
|
|
if (fs.existsSync(configPath)) {
|
|
config = JSON.parse(fs.readFileSync(configPath, 'utf-8'));
|
|
} else if (hasDefault) {
|
|
output(defaultValue, raw, String(defaultValue));
|
|
return;
|
|
} else {
|
|
error('No config.json found at ' + configPath);
|
|
}
|
|
} catch (err) {
|
|
if (err.message.startsWith('No config.json')) throw err;
|
|
error('Failed to read config.json: ' + err.message);
|
|
}
|
|
|
|
// Traverse dot-notation path (e.g., "workflow.auto_advance")
|
|
const keys = keyPath.split('.');
|
|
let current = config;
|
|
for (const key of keys) {
|
|
if (current === undefined || current === null || typeof current !== 'object') {
|
|
if (hasDefault) { output(defaultValue, raw, String(defaultValue)); return; }
|
|
error(`Key not found: ${keyPath}`);
|
|
}
|
|
current = current[key];
|
|
}
|
|
|
|
if (current === undefined) {
|
|
if (hasDefault) { output(defaultValue, raw, String(defaultValue)); return; }
|
|
error(`Key not found: ${keyPath}`);
|
|
}
|
|
|
|
output(current, raw, String(current));
|
|
}
|
|
|
|
/**
|
|
* Command to set the model profile in the config file.
|
|
*
|
|
* Note that this exits the process (via `output()`) even in the happy path.
|
|
*/
|
|
function cmdConfigSetModelProfile(cwd, profile, raw) {
|
|
if (!profile) {
|
|
error(`Usage: config-set-model-profile <${VALID_PROFILES.join('|')}>`);
|
|
}
|
|
|
|
const normalizedProfile = profile.toLowerCase().trim();
|
|
if (!VALID_PROFILES.includes(normalizedProfile)) {
|
|
error(`Invalid profile '${profile}'. Valid profiles: ${VALID_PROFILES.join(', ')}`);
|
|
}
|
|
|
|
// Ensure config exists (create if needed)
|
|
ensureConfigFile(cwd);
|
|
|
|
// Set the model profile in the config
|
|
const { previousValue } = setConfigValue(cwd, 'model_profile', normalizedProfile, raw);
|
|
const previousProfile = previousValue || 'balanced';
|
|
|
|
// Build result value / message and return
|
|
const agentToModelMap = getAgentToModelMapForProfile(normalizedProfile);
|
|
const result = {
|
|
updated: true,
|
|
profile: normalizedProfile,
|
|
previousProfile,
|
|
agentToModelMap,
|
|
};
|
|
const rawValue = getCmdConfigSetModelProfileResultMessage(
|
|
normalizedProfile,
|
|
previousProfile,
|
|
agentToModelMap
|
|
);
|
|
output(result, raw, rawValue);
|
|
}
|
|
|
|
/**
|
|
* Returns the message to display for the result of the `config-set-model-profile` command when
|
|
* displaying raw output.
|
|
*/
|
|
function getCmdConfigSetModelProfileResultMessage(
|
|
normalizedProfile,
|
|
previousProfile,
|
|
agentToModelMap
|
|
) {
|
|
const agentToModelTable = formatAgentToModelMapAsTable(agentToModelMap);
|
|
const didChange = previousProfile !== normalizedProfile;
|
|
const paragraphs = didChange
|
|
? [
|
|
`✓ Model profile set to: ${normalizedProfile} (was: ${previousProfile})`,
|
|
'Agents will now use:',
|
|
agentToModelTable,
|
|
'Next spawned agents will use the new profile.',
|
|
]
|
|
: [
|
|
`✓ Model profile is already set to: ${normalizedProfile}`,
|
|
'Agents are using:',
|
|
agentToModelTable,
|
|
];
|
|
return paragraphs.join('\n\n');
|
|
}
|
|
|
|
module.exports = {
|
|
VALID_CONFIG_KEYS,
|
|
cmdConfigEnsureSection,
|
|
cmdConfigSet,
|
|
cmdConfigGet,
|
|
cmdConfigSetModelProfile,
|
|
cmdConfigNewProject,
|
|
};
|