Skip to content
OpenAgentSkill
Browsing is open · Catalog captured 2026-11-21 · Automatic updates and account services are paused.
Development

claude-runner

Claude/Codex CLI execution layer — prompt loading, stream-json parsing, file output extraction, path sanitization, skill file writing, and skill rule generation

aspenkit · AI review on file

Installation

Read the source and dependencies before running a command. Copying does not install or execute anything.

npx skills add aspenkit/aspens --skill claude-runner

About this skill


name: claude-runner description: Claude/Codex CLI execution layer — prompt loading, stream-json parsing, file output extraction, path sanitization, skill file writing, and skill rule generation triggers: files: - src/lib/runner.js - src/lib/timeout.js - src/lib/skill-writer.js - src/lib/skill-reader.js - src/prompts/**/*.md keywords: - runClaude - runCodex - runLLM - stream-json - codex exec - parseFileOutput - sanitizePath - loadPrompt - writeSkillFiles - skill-rules - mergeSettings - resolveTimeout

You are working on the CLI execution layer — the bridge between assembled prompts and the claude -p / codex exec CLIs, plus skill file I/O.

Key Concepts

  • Stream-JSON protocol (Claude): runClaude() always passes --verbose --output-format stream-json. Output is NDJSON: type: 'result' has final text + usage; type: 'assistant' has text/tool_use blocks; type: 'user' has tool_result blocks.
  • JSONL protocol (Codex): runCodex() spawns codex exec --json --sandbox read-only --ephemeral. The --ask-for-approval never flag is conditionally included based on capability detection (see below). Prompt is passed via stdin ('-' placeholder arg) to avoid shell arg length limits. Stdin write happens after event handlers are attached so fast failures are captured. Events: item.completed/item.updated with normalized types.
  • Codex capability detection: getCodexExecCapabilities() (internal, cached) runs codex exec --help and checks if --ask-for-approval appears in the help text. Result is cached in module-level codexExecCapabilities variable. If the help check fails (e.g., codex not installed), capabilities default to { supportsAskForApproval: false }. runCodex() only adds --ask-for-approval never when supportsAskForApproval is true.
  • Unified routing: runLLM(prompt, options, backendId) is the shared entry point — dispatches to runClaude() or runCodex() based on backendId. Exported from runner.js so command handlers no longer need local routing helpers.
  • Codex internals (private): normalizeCodexItemType() converts PascalCase/kebab-case to snake_case. collectCodexText() recursively extracts text from nested event content. Both are internal to runner.js.
  • Prompt templating: loadPrompt(name, vars) resolves {{partial-name}} from src/prompts/partials/ first, then substitutes {{varName}} from vars. Target-specific vars (skillsDir, skillFilename, instructionsFile, configDir) are passed by command handlers.
  • File output parsing: Primary: <file path="...">content</file> XML tags. Fallback: <!-- file: path --> comment markers. parseFileOutput(output, allowedPaths) accepts optional { dirPrefixes, exactFiles } to override default allowed paths.
  • Path sanitization: sanitizePath(rawPath, allowedPaths) (internal) blocks .. traversal, absolute paths. Defaults: .claude/ prefix + CLAUDE.md exact. Multi-target callers pass expanded allowed paths via getAllowedPaths() from target.js.
  • Validation: validateSkillFiles() checks for truncation (XML tag collisions), missing frontmatter, missing sections, bad file path references.
  • Skill rules generation: extractRulesFromSkills() reads all skills via skill-reader.js, produces skill-rules.json (v2.0) with file patterns, keywords, and intent patterns.
  • Domain patterns: generateDomainPatterns() converts file patterns to bash detect_skill_domain() function using BEGIN/END markers.
  • Trigger parsing precedence: parseTriggersFrontmatter(content) returns { filePatterns, keywords, alwaysActivate } parsed from a triggers: block in YAML frontmatter (supports block lists, inline arrays, and alwaysActivate: true for the base skill); returns null when no triggers: key exists. parseActivationPatterns and parseKeywords prefer this frontmatter when present and fall back to legacy ## Activation / Keywords: line parsing for older skills.
  • Settings merge: mergeSettings() merges aspens hook config into existing settings.json. Detects aspens-managed hooks by ASPENS_HOOK_MARKERS (skill-activation-prompt, graph-context-prompt, post-tool-use-tracker, save-tokens-statusline, save-tokens-prompt-guard, save-tokens-precompact). Also handles statusLine merging — replaces existing statusLine only if the current one is aspens-managed (detected by isAspensHook), preserving user-custom statusLine configs. After merging hooks, dedupeAspensHookEntries() removes duplicate aspens-managed entries per event type.
  • Directory-scoped writes: writeTransformedFiles() handles files outside .claude/ (e.g., src/billing/AGENTS.md) with explicit path allowlist — only CLAUDE.md, AGENTS.md exact files and .claude/, .agents/, .codex/ prefixes are permitted.
  • findSkillFiles matching: Only matches the exact skillFilename (e.g., skill.md or SKILL.md), not arbitrary .md files in the skills directory.

Critical Rules

  • Both --verbose and --output-format stream-json are required for Claude — omitting either breaks stream parsing.
  • Codex uses --json --sandbox read-only --ephemeral — --sandbox read-only restricts filesystem access, --ephemeral avoids persisting conversation. --ask-for-approval never is added only if getCodexExecCapabilities() confirms support. Prompt goes via stdin, not as a CLI arg.
  • Codex stdin write order matters — event handlers (stdout, stderr, close, error) must be attached before writing to stdin, so fast failures are captured.
  • Path sanitization is non-negotiable — sanitizePath() blocks .. traversal, absolute paths, and any path not in the allowed set.
  • Prompt partials resolve before variables — {{skill-format}} resolves to partials/skill-format.md first. If no file, falls through to variable substitution.
  • **Timeout resolut
Read the complete source ↗