* Builds the finding description, appending function names if present.

* Builds the argument list for the Echidna process.

* Builds the argument list for the Halmos process.

* Builds the argument list for the Medusa process.

* Builds the request body for the Solodit API.

* Checks whether the `halmos` binary is available on PATH.

* Checks whether the `echidna` binary is available on PATH.

* Checks whether the `medusa` binary is available on PATH.

* Counts lines: \\n occurrences + 1, or 0 for empty content. * A trailing newline adds an extra line compared to `wc -l` (intentional per AC4).

* Ensures the work directory structure exists. * * Creates `<rootDir>/.sc-auditor-work/pocs/` if it does not already exist.

* Executes the Solodit API request with retry logic for 5xx errors.

* Extracts unique file paths from Aderyn instances.

* Extracts unique file paths from Slither elements.

Extract YAML frontmatter string from SKILL.md (between first two --- delimiters).

* Extracts function names from Slither elements.

* Extracts line range from Aderyn instances.

* Extracts line range from the first element's source_mapping.lines.

* Flattens tags from issues_issuetagscore[].tags_tag.title.

* Maps API impact (uppercase) to SoloditSeverity (capitalized).

* Generates structured exploit body from an exploit sketch. * * Produces labeled SETUP/ATTACK/ASSERTIONS sections populated from sketch fields.

* Handles HTTP error responses and returns appropriate error messages. * Returns null if the error should trigger a retry.

* Checks if cached data is still fresh (within TTL).

* Checks if stderr indicates a compilation error.

* Checks if stderr indicates a compilation error.

* Type guard for valid leaf items with all required fields.

* Loads a .env file from the config directory if it exists. * Sets environment variables that are NOT already set (env takes precedence). * Supports

* Parses Echidna text output into structured results.

* Parses Halmos text output into structured results.

* Parses Medusa text output into structured results.

* Parses rate limit reset timestamp into ISO date string.

* Reads and parses the Aderyn output file. * * Note: Unlike Slither, Aderyn does not include a 'success' field in its JSON output. * Aderyn's outpu

* Reads and parses the config file from disk. * Returns empty object if file doesn't exist (config.json is optional), * unless SC_AUDITOR_CONFIG was

* Reads and parses the Slither output file. * @returns Parsed output or error result.

* Resolves the config file path. * SC_AUDITOR_CONFIG env var takes precedence over the default path. * Returns whether the path was explicitly set v

* Runs the Aderyn subprocess.

* Runs the Echidna subprocess.

* Runs the Halmos subprocess.

* Runs the Medusa subprocess.

* Runs the Slither subprocess. * Note: Slither exits with non-zero when it finds issues, which is expected. * We only treat it as failure if the out

* Sanitizes a hotspot ID for use as a filename.

* Maps tool severity (capitalized) to API impact (uppercase).

* Validates and normalizes raw config input into a fully resolved Config object.

Recursively scan directory for leftover mcp__sc-auditor__ references.

* Validates and normalizes the llm_reasoning config section.

* Validates that an optional field, if present, is a number within [min, max].

* Validates and normalizes the proof_tools config section.

* Validates that report_output_dir, if present, is a non-empty relative path * without '..' traversal, absolute paths, or UNC paths.

* Validates that rootDir exists and is a directory.

* Validates and normalizes the static_analysis config section.

* Validates and normalizes the verify config section. * In benchmark mode, demote_unproven_medium_high defaults to true.

* Validates and normalizes the workflow config section.

* Recursively walks a directory, collecting .sol file paths. * Does not follow symlinks. Skips excluded and unreadable directories.

* Recursively walks the nested category tree and flattens it. * Attaches the parent category name to each leaf item. * Warns about malformed nodes t

* Writes data to cache atomically using temp file + rename. * This prevents concurrent readers from seeing partial files. * Uses a unique temp filen

* Converts a single Slither detector result into a Finding.

* Parses a single API issue into SoloditSearchResult.