Organize Cognitum Seed devices into fleets, or device groups. Create and list fleets, add or remove devices, and delete fleets through dedicated commands, with explicit confirmation required before deletion.
Analyze Git changes before submitting or merging a pull request. Assess risk and complexity, classify changes, suggest reviewer roles, and identify files needing closer inspection using diff statistics and file-level risk analysis.
Send a single request to DeepSeek for summarization, information extraction, or simple classification. Receive the response and token usage as JSON for further processing, with structured status information when the API is unavailable.
Inspect evolved prompt configurations, validate their structure, and view the system prompts they produce. You can also classify failures in run transcripts to support reviews before adopting or changing an agent policy.
Generate API documentation from TypeScript and JavaScript source code. Extract JSDoc comments, fill gaps in public API descriptions, and produce OpenAPI 3.0 definitions, a standard format for describing HTTP endpoints.
Run Ruflo background workers on a recurring schedule within a Claude Code session. Choose tasks such as security analysis, documentation, or test coverage analysis, check their status, and stop the loop after repeated reports of no new work.
Define reusable reasoning and decision patterns for agents, embed them in executable workflows, and review their performance. Share successful approaches, including conditional decision trees, stepwise problem solving, analogies, and evaluation from multiple perspectives.
Benchmark security detection systems against 10 known vulnerabilities and nine harmless decoys. Compare four approaches, including an evolved champion, and receive structured JSON covering detection, false alarms, patch tests, reproduction, unsafe outputs, and overall pass or fail.
Detect and classify unusual readings from Cognitum Seed devices, including spikes, flatlines, and drift. Review anomaly scores, recommend quarantine for scores above 0.9, and save detected patterns for future reference.
Schedule workers to run checks at fixed times while a Claude Code session stays open. Set up recurring security scans, test coverage analysis, or documentation tasks, and find or cancel scheduled jobs.
Investigate complex topics using web sources, stored knowledge, and relevant code. Break questions into smaller parts, compare evidence, and produce a structured report with confidence assessments, unresolved questions, and next steps.
Generate and update API or project documentation, and flag inconsistencies between existing docs and current code. Schedule recurring maintenance and save the documentation approach for reuse in later work.
Manage Git branches, resolve conflicts, and coordinate pull requests, which propose code changes. Check change risks and automated test status before merging, and inspect metrics such as merge frequency and review time.
Create a standard folder layout for a DDD bounded context, a module organized around one business area. Generate TypeScript export files, keep infrastructure outside the public exports, and record the context in a domain model graph.
Experiment with the policies that govern an AI agent, such as planning, memory, and tool use. Test one policy change per variant in a sandbox, an isolated environment, and promote only candidates with measured improvements.
Turn research findings stored in memory into structured reports with evidence ratings, source references, and practical recommendations. Compare conflicting claims, identify unresolved questions, and explain limitations so accumulated research can support decisions and next steps.
Explore a knowledge graph, a network of entities and relationships, from a chosen starting point. Rank relevant paths and summarize dependencies or indirect connections to help understand code structure and assess the impact of changes.
Compare saved benchmark runs over time to track win rate, response times, escalation rate, and LLM baseline costs. Review changes between the first and latest runs, with flags for declining win rates or rising average latency.
Prepare a node to participate in federation, a setup connecting multiple nodes. Generate an ed25519 keypair and configuration, optionally set a compliance mode, and display the public key and endpoint to share with peers.
Run three static audits covering architecture alignment, threats, and MCP server policies, permissions, and dependencies. Store timestamped results, track changes across runs, and fail scheduled CI jobs when findings reach a chosen severity.
Register a Cognitum Seed device at a supplied endpoint or its default USB Ethernet address. The workflow establishes an agent connection, retrieves device identity, assigns initial trust, optionally pairs the device, and records registration.
Manage firmware updates across IoT device fleets through staged rollouts. Check a small initial group for anomalies before expanding deployment, inspect rollout status, and roll back updates with explicit user approval for device changes.
Create a starter structure for a custom AI agent harness, the setup that supports agent execution. Choose a template and host, preview the plan, then explicitly confirm generation in a new directory outside the current project.
Review an AI team's work against its acceptance criteria using collected evidence. Match requirements to task results or artifacts, distinguish verified facts from agent claims, and identify uncertainty or missing work before recommending completion.
Is this your account? Link GitHub to bring it onto your creator page (Japanese)