Module runtime 

Purpose: Coordinate shared ProjectAtlas CLI and MCP runtime workflows. Shared runtime orchestration for the ProjectAtlas CLI and MCP adapters.

Modules

graph_projection 🔒

Normalize parser-owned symbol facts into one generation-bound repository graph.

module_resolution 🔒

Bounded repository compiler-configuration loading for semantic graph projection.

optional_parser_runtime 🔒

Process-wide admission and scheduling for the optional parser-pack worker.

source_observation 🔒

Process-local source observation for long-lived verified index reads.

Structs

DatabaseLintReport 🔒

Typed SQLite-owned portion of one lint result.

ExactFreshIndexRead 🔒

Current read snapshot established through exact source verification.

IndexFreshnessAssessment 🔒

Exact freshness assessment plus its measured source/database work.

IndexFreshnessDelta 🔒

Current local-source delta detected before a normal indexed read.

IndexInitRequired 🔒

Typed first-use handoff for one selected project root.

IndexProjectMismatch 🔒

Typed refusal when a selected project root and durable index disagree.

IndexPublicationBatch 🔒

Filesystem and derived facts prepared before acquiring the SQLite writer.

IndexPublicationOutcome 🔒

Reports produced after a staged batch commits successfully.

IndexRefreshReport 🔒

Combined refresh output for watcher and one-shot refresh paths.

IndexRefreshRequired 🔒

Bounded typed report returned before a stale indexed read can execute.

IndexVerificationIncomplete 🔒

Bounded typed report returned when source verification is incomplete.

InitBootstrapOptions 🔒

Options for the first-run initialization bootstrap.

InitHostConfigStatus 🔒

Status for one generated host integration config.

InitHydrationPhase 🔒

Registered-worktree initialization source state.

InitPathStatus 🔒

Status for one path managed by init.

InitScanPhase 🔒

Scan/index phase result for init.

InitSetupReport 🔒

First-run init report shared by CLI and MCP adapters.

LegacyPurposeReport 🔒

Legacy purpose cleanup report.

LintReport 🔒

Shared typed lint result consumed by CLI and MCP adapters.

OptionalParserSettingsReport 🔒

Optional parser state present in both feature-enabled and default-core-only builds.

PathStatus 🔒

Filesystem status for a diagnostic path.

ProjectWorktreeRequired 🔒

Typed refusal when a bare/common Git directory is selected as source.

PurposeCurationItem 🔒

One path that needs purpose curation.

PurposeCurationPage 🔒

Agent-facing purpose curation queue with bounded health metadata.

PurposeCuratorHandoff 🔒

Purpose curation handoff for agent/plugin harnesses.

PurposeImportLimits 🔒

Hard authored-purpose input limits for one publication attempt.

PurposeImportReport 🔒

Legacy purpose import counts from a scan.

PurposeImportSnapshot 🔒

Deterministic purpose-import rows and the inputs that produced them.

PurposeInputReader 🔒

Operation-owned reader for authored purpose and publication inputs.

PurposeReviewItem 🔒

Per-path batch review result.

PurposeReviewReport 🔒

Batch purpose review result.

PurposeReviewRequest 🔒

One agent-reviewed purpose update requested by a batch review.

ResetIndexReport 🔒

Runtime index/cache cleanup report.

ScanReport 🔒

Scan command report shared by CLI and MCP adapters.

ScanRuntimePlan 🔒

Resolved scan runtime policy shared by CLI and MCP adapters.

SettingsClassifiedNavigationReport 🔒

Content-free classified navigation capabilities shared by CLI and MCP settings.

SettingsIndexStats 🔒

Indexed state summary for settings diagnostics.

SettingsLexicalSearchReport 🔒

Typed lexical-search settings projection.

SettingsReport 🔒

Local settings report.

SettingsSearchModeReport 🔒

Typed settings projection for one search mode.

SettingsSearchReport 🔒

Content-free readiness for supported and planned search modes.

SourceVerificationWork 🔒

Measured work performed by one exact source-freshness verification.

StructuralSummaryDerivation 🔒

One file’s closed structural-summary derivation.

StructuralSummaryReport 🔒

Deterministic structural-summary refresh report.

StructuralSummaryStage 🔒

Structural summary mutations retained outside the SQLite writer transaction.

SymbolBuildOptions 🔒

Options controlling source parsing during symbol graph builds.

SymbolBuildReport 🔒

Symbol graph build report.

SymbolBuildStage 🔒

Symbol mutations retained outside the SQLite writer transaction.

SymbolParseJob 🔒

Source file queued for symbol parsing.

SymbolParseSuccess 🔒

Successful parser output waiting for sequential DB persistence.

SymbolPublicationLimits 🔒

Aggregate rows and retained string bytes admitted by one symbol publication.

TextIndexOptions 🔒

Options controlling full-text persistence for SQLite search.

TextIndexRefresh 🔒

Persisted text refresh result plus rows reused by structural summarizers.

TextIndexReport 🔒

Persisted file-text index report.

TextIndexRow 🔒

Outcome of considering one file for persisted text search.

UsageRuntimeInstance 🔒

One optional telemetry identity owned by a CLI invocation or MCP process.

WatchChangeSet 🔒

Debounced filesystem changes observed by watcher mode.

WatchReport 🔒

Watch command report.

WatchStatusReport 🔒

Watcher status report.

Enums

IndexReadStatus 🔒

Closed state returned when an index-backed read cannot proceed safely.

IndexRefreshReason 🔒

Closed reason for refusing an index-backed read.

IndexRefreshScope 🔒

Scope required to recover a stale index safely.

IndexVerificationReason 🔒

Closed reason why current local source could not be verified.

InitHydrationStatus 🔒

Stable worktree hydration outcomes exposed by init.

InitPhaseStatus 🔒

Project initialization phase status.

NodePublicationBatch 🔒

Node mutations owned by one full or incremental publication.

PurposeLintLevel 🔒

Purpose curation strictness used by projectatlas lint.

PurposeReviewAction 🔒

Stable purpose-review action values.

SettingsCapabilityState 🔒

Readiness of one settings capability.

SourceReadFailure 🔒

Failure from one cancellation-aware bounded source read.

StructuralSummaryChange 🔒

One closed structural-summary projection mutation.

SymbolParseOutcome 🔒

Outcome from one parser worker.

SymbolProjectionChange 🔒

One closed symbol projection mutation.

TextIndexSkipReason 🔒

Text-index outcome categories.

Constants

BUILTIN_PROJECTATLAS_PURPOSES 🔒

Built-in purposes for reserved project-local ProjectAtlas metadata inputs.

CONTROLLED_SOURCE_READ_BUFFER_BYTES 🔒

Chunk size used by cancellation-aware bounded source reads.

CORE_INDEX_POLICY_PATHS 🔒

Core project-local files whose edits can change source-selection policy.

DEFAULT_HEALTH_LIMIT 🔒

Default health rows returned when the caller does not request a page size.

DEFAULT_INDEX_WORK_TIMEOUT 🔒

Default whole-operation deadline when no narrower parser limit is supplied.

INDEX_DERIVATION_CONTRACT_VERSION 🔒

Explicit version of the built-in derived-index projection contract.

INDEX_FRESHNESS_SAMPLE_LIMIT 🔒

Maximum changed paths included in a freshness failure payload.

INDEX_WORKER_SAFE_CEILING 🔒

Maximum indexing workers regardless of a larger caller request.

MAX_HEALTH_LIMIT 🔒

Maximum health rows returned in one payload.

MAX_INCREMENTAL_CHANGED_PATHS 🔒

Maximum event paths accepted by one incremental watcher publication.

MAX_INCREMENTAL_SOURCE_BYTES 🔒

Maximum source bytes accepted by one incremental watcher publication.

MAX_INDEXED_NAVIGATION_SOURCE_BYTES 🔒

Maximum current source bytes one navigation read may allocate and inspect.

MAX_PUBLICATION_STAGING_BYTES 🔒

Maximum aggregate retained string bytes across one in-memory publication batch.

MAX_PURPOSE_HEADER_BYTES 🔒

Maximum source prefix inspected for a legacy purpose header.

MAX_PURPOSE_IMPORT_BYTES 🔒

Maximum aggregate authored-purpose bytes inspected by one publication.

MAX_PURPOSE_IMPORT_RECORDS 🔒

Maximum normalized legacy purpose rows admitted by one publication.

MAX_PURPOSE_INPUT_FILE_BYTES 🔒

Maximum complete config, map, or non-source purpose input size.

MAX_PURPOSE_REVIEW_FIELD_BYTES 🔒

Maximum bytes in one non-path purpose-review string field.

MAX_PURPOSE_REVIEW_INPUT_BYTES 🔒

Maximum aggregate string bytes admitted to one purpose-review batch.

MAX_PURPOSE_REVIEW_INPUT_FILE_BYTES 🔒

Maximum JSON bytes read for one CLI purpose-review batch.

MAX_PURPOSE_REVIEW_PATH_BYTES 🔒

Maximum bytes in one repository path supplied to purpose review.

MAX_PURPOSE_REVIEW_REPORT_BYTES 🔒

Maximum retained item/output bytes for one purpose-review report.

MAX_STAGED_TEXT_BYTES 🔒

Maximum UTF-8 source bytes retained while one publication is staged.

MAX_SYMBOL_FILE_BYTES 🔒

Maximum file size parsed for symbols by default.

MAX_SYMBOL_PARSE_JOBS 🔒

Maximum symbol candidates accepted by one publication.

NORMAL_READ_REFRESH_MAX_BYTES 🔒

Maximum current source bytes a normal read may reconcile before answering.

NORMAL_READ_REFRESH_MAX_PATHS 🔒

Maximum affected paths a normal read may reconcile before answering.

PUBLICATION_NODE_BATCH_SIZE 🔒

Maximum scan-node mutations applied between publication cancellation checks.

PUBLICATION_PATH_BATCH_SIZE 🔒

Maximum deleted repository paths applied between publication cancellation checks.

PUBLICATION_TEXT_BATCH_SIZE 🔒

Maximum persisted source texts applied between publication cancellation checks.

PURPOSE_CURATOR_RECOMMENDED_REASONING 🔒

Stable serialized recommendation for host-owned purpose curator selection.

PURPOSE_REVIEW_REPORT_ERROR_FIELD 🔒

Purpose-review report field name for an item error.

WATCH_EVENT_QUEUE_CAPACITY 🔒

Maximum native watcher events buffered before continuity becomes uncertain.

Functions

absolute_path 🔒

Return an absolute path without requiring the target to exist.

absolute_watch_path 🔒

Return an absolute path for a watcher event path.

admit_symbol_job_bytes 🔒

Read, bound, and hash-check one source exactly once for symbol staging.

apply_conditional_purpose_reviews 🔒

Apply one host-selected conditional batch with one database writer transaction.

apply_file_content_classification_stage 🔒

Apply staged file roles inside the parent generation transaction.

apply_purpose_import_snapshot 🔒

Apply staged legacy-purpose rows without overwriting current reviewed intent.

apply_structural_summary_stage 🔒

Apply prepared structural summaries inside the parent publication transaction.

apply_symbol_build_stage 🔒

Apply prepared symbol mutations inside the parent publication transaction.

apply_text_index_stage 🔒

Apply prepared persisted-text rows inside the parent publication transaction.

automatic_refresh_write_is_unavailable 🔒

Distinguish optional repair contention from database-integrity failures.

bounded_index_work_control 🔒

Apply the runtime’s safe whole-operation deadline without weakening caller bounds.

build_settings_report 🔒

Build settings diagnostics shared by CLI and MCP.

byte_count_to_tokens 🔒

Estimate source tokens from a searched byte count.

byte_size_to_tokens 🔒

Estimate source tokens from a byte count with the shared token heuristic.

canonical_project_root 🔒

Return a canonical absolute project root.

canonical_source_project_root 🔒

Return one exact source root from structural Git or non-Git evidence.

checked_symbol_publication_usage 🔒

Admit one aggregate symbol-publication resource before persistence.

classified_navigation_capabilities 🔒

Return the closed classified navigation contract without inspecting private data.

classified_ranked_file_nodes_with_reasons 🔒

Load ranked files with persisted classification and optional explicit selection.

clear_skipped_symbol_index 🔒

Clear stale symbol output while preserving structural summaries when present.

collect_purpose_reviews 🔒

Review one admitted item-oriented batch while bounding retained report data.

complete_index_publication 🔒

Commit only after the shared work boundary still permits publication.

conditional_purpose_review_action 🔒

Map database conditional-apply state into the stable review action contract.

config_candidates_for_root 🔒

Return supported config paths for one project root.

config_root_mismatch_error 🔒

Build the standard config/root mismatch error.

db_sidecar_path 🔒

Return the path to a SQLite sidecar file.

default_cli_project_root 🔒

Resolve the default CLI project root before opening an implicit database.

default_mcp_project_root 🔒

Resolve the default MCP project root without trusting the process cwd.

defaultable_cli_project_root 🔒

Resolve a CLI repository-root argument, using indexed state for the default ..

detect_index_freshness_controlled 🔒

Detect the complete local-source delta under one cooperative work boundary.

document_navigation_error 🔒

Map one document extraction failure to the navigation boundary’s typed error contract.

document_parse_error_outcome 🔒

Preserve typed cancellation/resource failures while surfacing malformed documents as input errors.

empty_index_refresh_report 🔒

Build the stable report for a verified watcher no-op.

empty_symbol_build_report 🔒

Return an empty symbol build report.

empty_text_index_report 🔒

Return an empty text-index report for a no-op refresh.

enforce_publication_staging_budget 🔒

Fail before writer acquisition when retained publication state exceeds its budget.

ensure_parent_dir 🔒

Create the parent directory for a path when it has one.

estimated_source_tokens_for_file_metadata 🔒

Estimate source tokens for persisted file metadata.

estimated_source_tokens_for_file_node 🔒

Estimate source tokens for one indexed file without reading it.

estimated_source_tokens_for_indexed_files 🔒

Estimate broad source tokens represented by indexed files with SQL aggregates.

estimated_source_tokens_for_path 🔒

Estimate source tokens for one indexed path, falling back safely for stale rows.

estimated_source_tokens_for_paths 🔒

Estimate source tokens for repository paths referenced by symbols/relations.

event_kind_affects_index 🔒

Return whether a notify event kind can change indexed content.

expected_nodes_after_incremental 🔒

Build the complete expected source state after one normalized incremental delta.

federated_worktree_error 🔒

Attach the exact alias to every typed or generic federated participant failure.

finish_purpose_import_fingerprint_controlled 🔒

Finish one purpose-import fingerprint from already normalized source records.

has_conditional_purpose_review_field 🔒

Return whether a row attempts to use queue-bound conditional review.

hash_index_contract_native_path 🔒

Add one native path to a fingerprint without a lossy text projection.

hash_index_contract_value 🔒

Add one unambiguous field/value pair to a derived-index fingerprint.

hash_publication_input_file_controlled 🔒

Bind one optional file’s identity and exact bytes to publication input state.

health_next_start_index 🔒

Return the next start index for a bounded health page.

health_severity_name 🔒

Return a stable lowercase severity name.

hydrate_purpose_review_classifications 🔒

Add registry-owned file roles without allowing purpose writes to alter them.

index_derivation_fingerprint 🔒

Hash the parser registry and source-selection policy that own derived rows.

index_derivation_fingerprint_for_contract 🔒

Hash one exact parser, semantic, and source-selection contract.

index_init_required 🔒

Return a typed, non-mutating first-use handoff for one selected root.

index_policy_path 🔒

Return whether one repository-relative path owns derived-index policy.

index_policy_refresh_required 🔒

Return the typed full-refresh state for a changed derivation contract.

index_work_control 🔒

Create one indexing boundary with the caller timeout capped by the safe default.

indexed_file_hashes 🔒

Load indexed file hashes for incremental refresh comparison.

indexed_file_hashes_for_paths 🔒

Load indexed file hashes for selected repository paths.

indexed_file_texts_for_nodes_controlled 🔒

Build bounded UTF-8 text rows while observing cancellation between files.

indexed_file_texts_for_nodes_with_limit 🔒

Build UTF-8 text rows under an explicit aggregate staging-byte limit.

indexed_native_path 🔒

Build an absolute native path for a previously validated indexed file key.

indexed_project_root 🔒

Load the project root recorded by the latest scan.

indexed_purpose_files 🔒

Collect .purpose files only from folders included in the normal index.

init_config_path 🔒

Return the config path init should preserve or create for a project root.

init_next_steps 🔒

Return concise next steps for humans and agents.

init_path_status 🔒

Return created/existing status for a path.

is_complete_conditional_purpose_review 🔒

Return whether a row carries one complete queue-bound conditional review.

is_generic_context_segment 🔒

Return whether a path segment is too generic to add useful purpose context.

is_gradle_build_script 🔒

Return whether a path is a Gradle build script rather than ordinary Kotlin/Groovy source.

is_symbol_candidate 🔒

Return whether a language should be parsed for symbols.

is_symbol_candidate_for_admission 🔒

Apply the project-selected optional-language boundary to symbol work admission.

join_human_names 🔒

Join declaration names as a compact human phrase.

legacy_project_root_candidate 🔒

Resolve a predecessor root only as a read-only discovery candidate.

lint_database_if_present 🔒

Build typed lint facts for an existing SQLite index.

lint_project 🔒

Build one shared typed lint result for CLI and MCP callers.

load_scan_import_config 🔒

Load map configuration for purpose import during scan.

load_synchronized_repository_token_report 🔒

Load one combined repository report while active worktree membership is stable.

lossless_native_path_display 🔒

Return a lossless UTF-8 display for one native path.

lossless_project_root_display 🔒

Return a lossless UTF-8 display for one existing canonical project root.

mcp_config_path_for_db 🔒

Return the project-local MCP config path associated with a database path.

move_reset_file_noclobber 🔒

Move one existing reset target without overwriting another path.

native_display_relative_path 🔒

Reconcile equivalent native paths when Windows extended prefixes differ.

next_step_report_payload 🔒

Build the flattened agent-facing next-step payload.

next_step_report_with_selection 🔒

Build a next-step report under one optional classified-content selection.

normalized_deleted_path 🔒

Normalize a deleted path if it belongs to the watched repository.

normalized_folder_filter 🔒

Normalize a folder filter into the repository path convention.

normalized_watch_index_path 🔒

Return one repository-contained native path after watcher normalization and policy checks.

notify_event_changes 🔒

Convert a notify event into index-relevant changes.

notify_result_changes 🔒

Convert a notify result into index-relevant changes.

notify_runtime_available 🔒

Return whether the platform watcher can be constructed in this process.

observed_content_noun 🔒

Return the truthful noun for registry-classified source versus documentation.

observed_language_label 🔒

Return a readable language label for agent-facing content summaries.

observed_manifest_subject 🔒

Return the subject phrase for manifest-style content summaries.

open_atlas_store_for_project 🔒

Open or create a durable index bound to one selected project root.

open_atlas_store_for_project_with_location_validator 🔒

Validate storage before creating the database parent and opening the store.

open_atlas_store_read_only_for_project 🔒

Open a current durable index read snapshot bound to one selected root.

open_exact_fresh_atlas_store_for_project_controlled 🔒

Open a current snapshot and retain exact freshness work for epoch accounting.

open_exact_fresh_atlas_store_for_project_with_repair 🔒

Open a current read snapshot without repairing stale source or durable state.

open_exact_saved_source_matches_index_controlled 🔒

Open an exact current snapshot without publishing a source repair.

open_federated_atlas_stores_for_project 🔒

Open an explicit ordered set of current project indexes without mutating any root.

open_fresh_atlas_store_for_project 🔒

Open a current read snapshot, reconciling one safe bounded delta when possible.

open_fresh_atlas_store_for_project_controlled 🔒

Open a current read snapshot under one cooperative freshness boundary.

parse_admitted_symbol_job 🔒

Extract conservative facts from admitted source and retain independent source provenance.

parse_document_symbol_job 🔒

Parse one admitted PDF/DOCX byte stream into the existing publication shape.

parse_symbol_job_controlled 🔒

Parse one source file while observing cancellation before and during parsing.

path_context_subject 🔒

Return a path-aware subject phrase for a generated purpose suggestion.

path_segment_words 🔒

Convert one path segment into readable lowercase words.

path_status 🔒

Return a diagnostic status for one path.

preflight_existing_project_binding 🔒

Refuse an existing database binding before init writes.

preview_conditional_purpose_review 🔒

Preview one conditional review against the current queue state without writing.

primary_symbol_kinds 🔒

Return a compact phrase describing the most important symbol kinds.

primary_symbol_names 🔒

Return stable names for the most important declaration symbols.

project_root_from_db_path 🔒

Infer a project root from a default .projectatlas/projectatlas.db path.

project_store_error 🔒

Preserve typed selected-root mismatch diagnostics across store adapters.

protected_purpose_paths 🔒

Return paths whose reviewed or built-in purpose must suppress generated suggestions.

publication_base_generation 🔒

Capture the last complete generation used as a publication compare-and-swap anchor.

publication_contract_matches 🔒

Return whether symbol reuse and incremental publication share the current derivation contract.

publication_input_error 🔒

Preserve typed work failures while adapting authored-input uncertainty.

publish_index_batch 🔒

Apply one fully prepared index batch in a short generation-checked transaction.

purpose_curation_page 🔒

Build a purpose curation queue from the bounded health page.

purpose_curator_handoff 🔒

Build one host-owned purpose-curator handoff shared by init and session brief.

purpose_handoff_instructions 🔒

Return stable purpose handoff instructions for agent harnesses.

purpose_header_candidates 🔒

Return source files that appear to start with legacy Purpose headers.

purpose_queue_curation_scope 🔒

Return the truthful curation tier selected by explicit queue scope flags.

purpose_queue_file_scope 🔒

Return the file inclusion scope for purpose curation metadata.

purpose_queue_folder_scope 🔒

Return the folder inclusion scope for purpose curation metadata.

purpose_review_input_too_large 🔒

Build the stable aggregate-byte admission failure.

purpose_review_item_bytes 🔒

Return retained string bytes for one report row after per-field admission.

purpose_review_report_too_large 🔒

Build the stable purpose-review report/output limit failure.

push_word_space 🔒

Append one word separator when the phrase already has content.

ranked_file_nodes_with_reasons 🔒

Load ranked file nodes with concise reasons.

ranked_folder_nodes_with_reasons 🔒

Load ranked folder nodes with concise reasons.

read_indexed_file_content 🔒

Read content for a previously validated indexed file key.

read_source_bytes_controlled 🔒

Read at most one admitted source-byte budget while checking cooperative stop state.

reconcile_hydrated_index_controlled 🔒

Reconcile a copied worktree baseline through exact no-op, incremental, or full refresh.

record_directory_walk_usage_estimate 🔒

Record a broad directory-walk avoidance estimate.

record_usage_estimate 🔒

Record a usage event from a fast baseline estimate and actual atlas payload.

record_usage_estimate_with_context 🔒

Record a usage event from a fast baseline estimate and explicit baseline semantics.

record_usage_text 🔒

Record a usage event from baseline and emitted text unless telemetry is disabled.

refresh_index 🔒

Refresh filesystem and symbol state.

refresh_index_controlled 🔒

Refresh every derived projection under one cancellation boundary.

refresh_index_for_changes 🔒

Refresh filesystem and symbol state for a debounced event batch.

refresh_index_for_changes_controlled 🔒

Refresh one watcher batch under one cancellation and publication boundary.

relation_targets 🔒

Return relation targets for one relation kind.

render_classified_ranked_file_rows 🔒

Preserve existing ranked file rows while adding their persisted classification.

render_classified_symbol_rows 🔒

Preserve existing symbol rows while adding their owning file classification.

render_coverage_report 🔒

Render one bounded current coverage page as compact TOON.

render_health_page 🔒

Render a bounded health page as compact TOON.

render_purpose_curation_page 🔒

Render a purpose curation queue as compact TOON.

render_purpose_review_report 🔒

Render a batch purpose review report as compact TOON.

repository_path_is_absent 🔒

Match an exact absent repository key or one of its slash-delimited ancestors.

require_current_worktree_usage_snapshot 🔒

Require the atlas currently published at this path to match a captured snapshot exactly.

require_registered_worktree_lifecycle 🔒

Rediscover one registration immediately before a lifecycle-sensitive catalog write.

require_synchronization_control_identity 🔒

Keep synchronization on the control atlas captured by an alias-routed request.

reset_index_files 🔒

Preview or remove local runtime index/cache files.

reset_index_files_with_revalidation 🔒

Remove registered reset targets only after staging and lifecycle revalidation.

reset_index_targets 🔒

Return the fixed file inventory owned by one index reset.

resolved_mcp_config_path 🔒

Resolve the config path that should travel with generated MCP configs.

restore_staged_reset_files 🔒

Restore staged reset files without overwriting a replacement lifecycle.

revalidate_staged_publication_inputs_controlled 🔒

Recheck policy and exact source against one off-writer publication batch.

revalidate_staged_publication_inputs_controlled_with_limits 🔒

Recheck a staged batch under explicit purpose-input limits used by tests.

revalidate_staged_publication_inputs_with_purpose_snapshot 🔒

Recheck a full scan while reusing purpose rows from exact unchanged source nodes.

review_purpose_request 🔒

Validate and optionally apply one agent-reviewed purpose record.

review_purposes 🔒

Validate and optionally apply a batch of agent-reviewed purpose records.

run_init_bootstrap 🔒

Run the one-call first-run init bootstrap.

run_notify_watch_loop 🔒

Run an event-backed watcher loop with notify.

run_polling_watch_loop 🔒

Run the portable polling watcher fallback loop.

run_scan_pipeline_controlled 🔒

Execute the full pipeline under one cancellation and resource boundary.

run_single_watch_refresh 🔒

Run one deterministic watcher refresh and exit.

run_single_watch_refresh_controlled 🔒

Run one watcher refresh under one cancellation and publication boundary.

run_symbol_build_pipeline_controlled 🔒

Rebuild symbol projections under one cancellation and publication boundary.

run_watch_loop 🔒

Run the watcher refresh loop.

run_watch_with_polling_fallback 🔒

Run an event-backed watcher and preserve current changes through polling fallback.

safe_watch_relative_path 🔒

Return a safe normalized repository path for a watcher event.

same_indexed_source 🔒

Compare source-derived node identity while ignoring non-semantic mtimes.

scan_options_for_root 🔒

Load scan options for a project root from ProjectAtlas config when present.

seed_builtin_projectatlas_purposes 🔒

Seed built-in purposes for reserved ProjectAtlas metadata nodes when needed.

selected_scan_import_config_path 🔒

Resolve the exact config file selected for a scan plan without loading it.

settings_index_stats 🔒

Build index statistics for settings diagnostics.

settings_publication_matches 🔒

Reject mixed settings projections when publication changed between read snapshots.

sorted_watch_paths 🔒

Return event paths in deterministic order.

source_changed_during_derivation 🔒

Refuse publication when source bytes no longer match their staged node.

source_input_byte_limit 🔒

Keep document admission at its parser-owned ceiling and ordinary source at caller policy.

source_inspection_error 🔒

Preserve typed work failures while adapting ordinary scan uncertainty.

source_node_delta 🔒

Build one deterministic affected-path plan from current and indexed nodes.

stage_file_content_classifications 🔒

Derive one registry-owned or bounded-text fallback role per staged file.

stage_full_index_publication 🔒

Prepare a complete source/index batch without acquiring the SQLite writer.

stage_structural_summaries_for_nodes_controlled 🔒

Derive structural summary mutations without acquiring the SQLite writer.

stage_symbols_for_nodes_with_limits 🔒

Build selected symbol mutations without acquiring the SQLite writer.

stage_text_index_for_changed_paths_controlled 🔒

Build selected persisted-text rows without acquiring the SQLite writer.

staged_file_content_classification_bytes 🔒

Count retained classification paths and stable enum spellings.

staged_full_refresh_is_unchanged 🔒

Return whether one fully staged watcher refresh matches complete durable state.

staged_node_bytes 🔒

Count retained node string bytes for one bounded in-memory publication batch.

staged_publication_identity_bytes 🔒

Count the selected root and derivation identity retained by one batch.

staged_purpose_bytes 🔒

Count retained legacy-purpose strings for one staged batch.

staged_string_bytes 🔒

Count retained strings duplicated into one publication batch.

staged_text_bytes 🔒

Count retained persisted-text strings for one staged batch.

standalone_index_work_control 🔒

Create a bounded work boundary for runtime paths without symbol options.

strip_declaration_kind_prefix 🔒

Remove the deterministic symbol-kind phrase before the primary names.

strip_legacy_purpose 🔒

Inspect and optionally remove legacy .purpose files.

suggest_file_purpose 🔒

Create a generated file-purpose suggestion from a path and content summary.

summarize_purpose_review 🔒

Aggregate stable batch counters from per-path review outcomes.

summarize_symbol_graph 🔒

Create a deterministic one-line content summary from extracted symbols.

summary_between 🔒

Return a non-empty substring between two markers.

summary_primary_declarations 🔒

Extract primary declaration names from a deterministic content summary.

symbol_limits_with_remaining_staging_bytes 🔒

Restrict parser output to the remaining aggregate publication-staging budget.

symbol_parse_output_bytes 🔒

Count retained string bytes in one parser output without serializing a second copy.

synchronize_registered_worktree_usage_with_catalog_validation 🔒

Synchronize active registrations with one injectable final catalog boundary.

telemetry_disabled 🔒

Return whether telemetry writes are disabled for read-only review contexts.

text_index_options 🔒

Resolve text-index persistence options from command override and config.

trim_summary_clause 🔒

Trim trailing summary details from a declaration clause.

valid_watch_relative_path 🔒

Reject empty, current-directory, and parent traversal path components.

validate_purpose_review_admission 🔒

Enforce shared CLI/MCP purpose-review request limits before database work.

validate_purpose_review_field 🔒

Validate one caller-controlled purpose-review string before retaining output.

validate_purpose_review_report 🔒

Enforce exact supported adapter output caps for one completed report.

validated_file_key 🔒

Normalize and validate a user-supplied path as a repository-relative file key.

validated_indexed_file_key 🔒

Validate that a path belongs to the indexed project file set.

value_only_symbol_kind_label 🔒

Return the right value-only summary noun for the indexed language.

verification_incomplete 🔒

Convert a policy/root preflight failure into a non-destructive read refusal.

verify_index_project_root 🔒

Verify that the opened database belongs to the selected canonical root.

verify_index_publication 🔒

Reject mixed or runtime-incompatible derived projections before source reads.

verify_saved_source_matches_index_controlled 🔒

Verify exact saved source against the index without publishing a repair.

verify_source_nodes_match 🔒

Compare a current exact scan with the source-derived nodes being validated.

wait_for_index_event_with_continuity 🔒

Wait for one bounded event batch and preserve local queue-overflow uncertainty.

watch_path_requires_full_scan 🔒

Return whether a path event requires a full scan for correctness.

watcher_status_report 🔒

Build a watcher status report from a lightweight runtime probe.

with_synchronized_registered_worktree_usage 🔒

Synchronize active registrations and read the aggregate before catalog writers resume.

worker_count_for_work 🔒

Bound a worker pool by its work cardinality and runtime ceiling.