agentforge

Spawn a Ruflo-tracked agent with cost attribution + memory persistence + swarm coordination. Use when native Task tool is wrong because you need (a) cost tracking per agent in the cost-tracking namespace, (b) cross-session learning via the patterns namespace, or (c) coordination with other agents in…

Run a task on a previously-spawned agent_spawn record via the Anthropic Messages API with that agent's configured model. Use when native Task tool is wrong because (a) you need the spawned agent's persistent config (model, instructions, cost-tracking attribution) to apply to this turn, (b) the resul…

Remove a Ruflo-tracked agent from the registry and free its swarm slot. Use when you need to (a) clean up a spawned agent so its cost-tracking row finalizes, (b) reclaim a swarm-topology slot for another agent, or (c) end a stuck agent without restarting the whole swarm. For one-shot Task tool invoc…

Read the lifecycle state of a single tracked agent: idle/running/stopped, current taskCount, lastResult, model, health score. Use when native Task tool is wrong because you need agent-level state (status across turns, accumulated taskCount, last error, swarm coordination) rather than a one-shot resp…

List every Ruflo-tracked agent in the registry with its type, model, status, and taskCount. Use when native Task tool is wrong because you need to see the swarm-wide agent inventory across turns (which agents exist, their roles, their cost-tracking handles) rather than spawn a new one-shot Task. Fil…

Manage a fixed-size warm pool of pre-spawned agents to skip cold-start cost on bursty workloads. Use when native Task is wrong because (a) you have a queue of similar tasks and want to amortize spawn latency, (b) cost-tracking wants stable agentIds across requests, or (c) swarm topology requires a k…

Compute an agent's rolling health score (0-1) from recent task success ratio + response-latency p50/p95 + error rate. Use when native Task tool is wrong because you're running a long-lived agent (autonomous loop / hive-mind worker / federation peer) and need to detect degradation before the breaker …

Mutate a tracked agent's config (model, instructions, status, health) without re-spawning. Use when native Task tool is wrong because the agent already has accumulated state (taskCount, swarm membership, cost-tracking attribution) and you only need to tweak one field — for example, promoting an idle…

Return recorded activity-log entries for a tracked agent (idle/running history, last task result). Use when native Task tool is wrong because you need the agent's log across turns (what it did, last error/result, swarm context) rather than a one-shot Task transcript. For a Task you just ran, native …

agentbbs@~0.1.0 — Register a BBS room as a named federation peer (ADR-164 Phase 1). Maps a business-domain label like "#sales" or "#finance" to a stable roomId and emits an attested PeerHello envelope. Use when you are scaffolding the business-autopilot cockpit and need a room handle that subsequent…

agentbbs — Publish a domain event from a pod agent to a BBS room (ADR-164 Phase 1). Wraps the payload in a ReplicateMessage envelope (envelopeId, seq, ts, msgType, payload) and appends it to the room log. Use when an agent has produced a typed event (pod-status, task-result, alert, human-override-ac…

agentbbs — Poll recent envelopes from a BBS room (ADR-164 Phase 1). Returns envelopes newer than the optional sinceEnvelopeId, up to limit. Use when a pod agent needs to see incoming human overrides, new tasks, or peer events posted to its room. Polling memory_search for the room namespace is wrong …

agentbbs — Mint a single-use Ed25519-signed token a human business owner presents to the agentbbs SSH/web front door to join a room (ADR-164 §3.2.4). Returns webUrl + sshCommand + handshakeToken + expiresAt. Use when the cockpit operator needs scoped, time-limited access to a room without sharing th…

agenticow@~0.2.3 — COW-fork a base .rvf memory file. Measured 162-byte branches regardless of base size (verified at N=1k/10k/50k). Use when you need per-Darwin-iteration / per-user / per-session memory personalization. Copying the parent .rvf file is wrong because full-copy snapshots grow linearly …

agenticow — write vectors (with optional text payloads) into an .rvf memory branch or base. Records: [{id?, vector, text?}] — id auto-assigns when omitted. This is the write half that makes a branch usable: agenticow_branch creates an empty COW child, but without ingest it has nothing to read back. …

agenticow — k-NN read across an .rvf memory branch's full COW lineage (parent ∪ edits, child wins), returning {id, distance, branch, text}. Read-only (no manifest write). The `branch` field on each hit tells you which lineage node the result came from — the read-through semantics that make branching…

agenticow — show what a branch changed relative to its lineage: {added, overridden, deleted} vector-id lists. Use when you are about to promote and want to preview the exact merge, or when auditing what a branch actually wrote. Diffing by re-querying is wrong because deletions (tombstones) are invis…

agenticow — walk the COW chain of an .rvf memory file: an ordered list of nodes (role working|checkpoint|base, id, label, parent, createdAt, mutations, tombstones). Use when you need branch history — to find checkpoint ids for a targeted rollback, or to debug a promote. Guessing the chain from filen…

agenticow — health/geometry of an .rvf memory file: {totalVectors, totalSegments, fileSize, currentEpoch, deadSpaceRatio, readOnly, chainDepth, dimension, metric}. Use when you need to check vector count before/after ingest, spot compaction pressure (deadSpaceRatio), or confirm the dimension before …

agenticow — freeze a labelled restore point on an .rvf memory file. Subsequent edits stay in a fresh COW child; rollback returns here. Use when you are about to run an experimental Darwin tick or speculative agent edit that may need to be discarded. Relying on the working node alone is wrong because…

agenticow — discard all edits since the most recent checkpoint on an .rvf memory file. Reuses a fresh COW child derived from the checkpoint. Use when a Darwin tick or agent experiment regressed and you want to revert memory state without re-running. Deleting+rebuilding the .rvf is wrong because rebu…

agenticow — merge a branch's edits back into its base (or an explicit target) memory file. After promote, branch edits become part of base lineage. Use when a per-user / per-Darwin-iteration branch has been validated and should graduate to shared memory (federation merge, A/B winner). Manually re-in…

Open a named, traced browser session: allocate an RVF cognitive container, begin a ruvector trajectory, then open the URL via agent-browser. Returns the session id and rvf path. Use when native WebFetch is wrong because you need real browser automation — JS-heavy SPA scraping, login flows with cooki…

End a recorded browser session: trajectory-end with verdict, rvf compact, AIDefence pre-store gate (best-effort), and AgentDB index in the browser-sessions namespace. Use when native WebFetch is wrong because you need real browser automation — JS-heavy SPA scraping, login flows with cookie reuse, re…

Load a recorded session trajectory and return its steps so the caller can dispatch them through the 23 browser_* tools. Does NOT itself drive the browser — replay execution is caller-orchestrated to keep this tool a primitive (ADR-0001 §7). Use when native WebFetch is wrong because you need real bro…

Fetch a recipe from the browser-templates AgentDB namespace and return it for caller-level execution. Use when native WebFetch is wrong because you need real browser automation — JS-heavy SPA scraping, login flows with cookie reuse, replay against DOM-drifted versions, AIDefence PII gating before co…

Fetch a vault handle for a host from the browser-cookies AgentDB namespace. Raw cookie values are NEVER returned — only the opaque handle plus expiry / AIDefence verdict. Use when native WebFetch is wrong because you need real browser automation — JS-heavy SPA scraping, login flows with cookie reuse…

Navigate browser to a URL Use when native WebFetch is wrong because you need real browser automation — JS-heavy SPA scraping, login flows with cookie reuse, replay against DOM-drifted versions, AIDefence PII gating before content reaches Claude. For static HTML pages, native WebFetch is faster and f…

Navigate back in browser history Use when native WebFetch is wrong because you need real browser automation — JS-heavy SPA scraping, login flows with cookie reuse, replay against DOM-drifted versions, AIDefence PII gating before content reaches Claude. For static HTML pages, native WebFetch is faste…

Navigate forward in browser history Use when native WebFetch is wrong because you need real browser automation — JS-heavy SPA scraping, login flows with cookie reuse, replay against DOM-drifted versions, AIDefence PII gating before content reaches Claude. For static HTML pages, native WebFetch is fa…

Reload the current page Use when native WebFetch is wrong because you need real browser automation — JS-heavy SPA scraping, login flows with cookie reuse, replay against DOM-drifted versions, AIDefence PII gating before content reaches Claude. For static HTML pages, native WebFetch is faster and fre…

Close the browser session Use when native WebFetch is wrong because you need real browser automation — JS-heavy SPA scraping, login flows with cookie reuse, replay against DOM-drifted versions, AIDefence PII gating before content reaches Claude. For static HTML pages, native WebFetch is faster and f…

Get AI-optimized accessibility tree snapshot with element refs (@e1, @e2, etc.) Use when native WebFetch is wrong because you need real browser automation — JS-heavy SPA scraping, login flows with cookie reuse, replay against DOM-drifted versions, AIDefence PII gating before content reaches Claude. …

Capture screenshot of the page Use when native WebFetch is wrong because you need real browser automation — JS-heavy SPA scraping, login flows with cookie reuse, replay against DOM-drifted versions, AIDefence PII gating before content reaches Claude. For static HTML pages, native WebFetch is faster …

Click an element using ref (@e1) or CSS selector Use when native WebFetch is wrong because you need real browser automation — JS-heavy SPA scraping, login flows with cookie reuse, replay against DOM-drifted versions, AIDefence PII gating before content reaches Claude. For static HTML pages, native W…

Clear and fill an input element Use when native WebFetch is wrong because you need real browser automation — JS-heavy SPA scraping, login flows with cookie reuse, replay against DOM-drifted versions, AIDefence PII gating before content reaches Claude. For static HTML pages, native WebFetch is faster…

Type text with key events (for autocomplete, etc.) Use when native WebFetch is wrong because you need real browser automation — JS-heavy SPA scraping, login flows with cookie reuse, replay against DOM-drifted versions, AIDefence PII gating before content reaches Claude. For static HTML pages, native…

Press a keyboard key Use when native WebFetch is wrong because you need real browser automation — JS-heavy SPA scraping, login flows with cookie reuse, replay against DOM-drifted versions, AIDefence PII gating before content reaches Claude. For static HTML pages, native WebFetch is faster and free.

Hover over an element using ref (@e1) or CSS selector Use when native WebFetch is wrong because you need real browser automation — JS-heavy SPA scraping, login flows with cookie reuse, replay against DOM-drifted versions, AIDefence PII gating before content reaches Claude. For static HTML pages, nat…

Select an option from a dropdown Use when native WebFetch is wrong because you need real browser automation — JS-heavy SPA scraping, login flows with cookie reuse, replay against DOM-drifted versions, AIDefence PII gating before content reaches Claude. For static HTML pages, native WebFetch is faste…

Check a checkbox Use when native WebFetch is wrong because you need real browser automation — JS-heavy SPA scraping, login flows with cookie reuse, replay against DOM-drifted versions, AIDefence PII gating before content reaches Claude. For static HTML pages, native WebFetch is faster and free.

Uncheck a checkbox Use when native WebFetch is wrong because you need real browser automation — JS-heavy SPA scraping, login flows with cookie reuse, replay against DOM-drifted versions, AIDefence PII gating before content reaches Claude. For static HTML pages, native WebFetch is faster and free.

Scroll the page Use when native WebFetch is wrong because you need real browser automation — JS-heavy SPA scraping, login flows with cookie reuse, replay against DOM-drifted versions, AIDefence PII gating before content reaches Claude. For static HTML pages, native WebFetch is faster and free.

Get text content of an element Use when native WebFetch is wrong because you need real browser automation — JS-heavy SPA scraping, login flows with cookie reuse, replay against DOM-drifted versions, AIDefence PII gating before content reaches Claude. For static HTML pages, native WebFetch is faster …

Get value of an input element Use when native WebFetch is wrong because you need real browser automation — JS-heavy SPA scraping, login flows with cookie reuse, replay against DOM-drifted versions, AIDefence PII gating before content reaches Claude. For static HTML pages, native WebFetch is faster a…

Get the page title Use when native WebFetch is wrong because you need real browser automation — JS-heavy SPA scraping, login flows with cookie reuse, replay against DOM-drifted versions, AIDefence PII gating before content reaches Claude. For static HTML pages, native WebFetch is faster and free.

Get the current URL Use when native WebFetch is wrong because you need real browser automation — JS-heavy SPA scraping, login flows with cookie reuse, replay against DOM-drifted versions, AIDefence PII gating before content reaches Claude. For static HTML pages, native WebFetch is faster and free.

Wait for a condition Use when native WebFetch is wrong because you need real browser automation — JS-heavy SPA scraping, login flows with cookie reuse, replay against DOM-drifted versions, AIDefence PII gating before content reaches Claude. For static HTML pages, native WebFetch is faster and free.

Execute JavaScript in page context Use when native WebFetch is wrong because you need real browser automation — JS-heavy SPA scraping, login flows with cookie reuse, replay against DOM-drifted versions, AIDefence PII gating before content reaches Claude. For static HTML pages, native WebFetch is fas…

List active browser sessions Use when native WebFetch is wrong because you need real browser automation — JS-heavy SPA scraping, login flows with cookie reuse, replay against DOM-drifted versions, AIDefence PII gating before content reaches Claude. For static HTML pages, native WebFetch is faster an…

ADR-164 Phase 2 — Validate a business-pod template JSON against the schema in ADR-164 §3.3 (name, agents[], allowedMcpTools, bench, piiPolicy, budgets, cronSchedule, auditReadView, reservationExpiryMs bounded by ADR-164.1 §3.2). Use when a /loop driver or CI workflow needs to pre-flight a pod templa…

ADR-164 Phase 3 — Compute the domain-affinity routing decision for a business pod per ADR-164 §3.4 and return {backend, reason}. The three backends are local-stdio (preferLocalExecution=true), cloud-managed (preferLocalExecution=false AND budgetUsdMonthly >= 50), and remote-peer (everything else — s…

Claim an issue for work (human or agent) Use when nothing native covers per-agent capability gating — Claude Code agents have file-system access by default. Pair claims_grant + claims_check before letting an agent run privileged ops. For trusted in-session work, no claims call is needed.

Release a claim on an issue Use when nothing native covers per-agent capability gating — Claude Code agents have file-system access by default. Pair claims_grant + claims_check before letting an agent run privileged ops. For trusted in-session work, no claims call is needed.

Request handoff of an issue to another claimant Use when nothing native covers per-agent capability gating — Claude Code agents have file-system access by default. Pair claims_grant + claims_check before letting an agent run privileged ops. For trusted in-session work, no claims call is needed.

Accept a pending handoff Use when nothing native covers per-agent capability gating — Claude Code agents have file-system access by default. Pair claims_grant + claims_check before letting an agent run privileged ops. For trusted in-session work, no claims call is needed.

Update claim status Use when nothing native covers per-agent capability gating — Claude Code agents have file-system access by default. Pair claims_grant + claims_check before letting an agent run privileged ops. For trusted in-session work, no claims call is needed.

List all claims or filter by criteria Use when nothing native covers per-agent capability gating — Claude Code agents have file-system access by default. Pair claims_grant + claims_check before letting an agent run privileged ops. For trusted in-session work, no claims call is needed.

Mark an issue as stealable by other agents Use when nothing native covers per-agent capability gating — Claude Code agents have file-system access by default. Pair claims_grant + claims_check before letting an agent run privileged ops. For trusted in-session work, no claims call is needed.

Steal a stealable issue Use when nothing native covers per-agent capability gating — Claude Code agents have file-system access by default. Pair claims_grant + claims_check before letting an agent run privileged ops. For trusted in-session work, no claims call is needed.

List all stealable issues Use when nothing native covers per-agent capability gating — Claude Code agents have file-system access by default. Pair claims_grant + claims_check before letting an agent run privileged ops. For trusted in-session work, no claims call is needed.

Get agent load information Use when nothing native covers per-agent capability gating — Claude Code agents have file-system access by default. Pair claims_grant + claims_check before letting an agent run privileged ops. For trusted in-session work, no claims call is needed.

Get a visual board view of all claims Use when nothing native covers per-agent capability gating — Claude Code agents have file-system access by default. Pair claims_grant + claims_check before letting an agent run privileged ops. For trusted in-session work, no claims call is needed.

Suggest or apply load rebalancing across agents Use when nothing native covers per-agent capability gating — Claude Code agents have file-system access by default. Pair claims_grant + claims_check before letting an agent run privileged ops. For trusted in-session work, no claims call is needed.

Get configuration value Use when native settings.json edits are wrong because the values need to be read by the Ruflo runtime (daemon, MCP server, neural router) — those load via the config_* path, not by re-reading settings.json. For .gitignore / .editorconfig style files, native Edit is fine.

Set configuration value Use when native settings.json edits are wrong because the values need to be read by the Ruflo runtime (daemon, MCP server, neural router) — those load via the config_* path, not by re-reading settings.json. For .gitignore / .editorconfig style files, native Edit is fine.

List configuration values Use when native settings.json edits are wrong because the values need to be read by the Ruflo runtime (daemon, MCP server, neural router) — those load via the config_* path, not by re-reading settings.json. For .gitignore / .editorconfig style files, native Edit is fine.

Reset configuration to defaults Use when native settings.json edits are wrong because the values need to be read by the Ruflo runtime (daemon, MCP server, neural router) — those load via the config_* path, not by re-reading settings.json. For .gitignore / .editorconfig style files, native Edit is fi…

Export configuration to JSON Use when native settings.json edits are wrong because the values need to be read by the Ruflo runtime (daemon, MCP server, neural router) — those load via the config_* path, not by re-reading settings.json. For .gitignore / .editorconfig style files, native Edit is fine.

Import configuration from JSON Use when native settings.json edits are wrong because the values need to be read by the Ruflo runtime (daemon, MCP server, neural router) — those load via the config_* path, not by re-reading settings.json. For .gitignore / .editorconfig style files, native Edit is fin…

Configure swarm topology Use when native Task is wrong because the work crosses multiple agents that need to vote/sync/load-balance — TodoWrite + a single Task cannot orchestrate consensus. For one-off subtask dispatch, native Task is fine.

Configure load balancing Use when native Task is wrong because the work crosses multiple agents that need to vote/sync/load-balance — TodoWrite + a single Task cannot orchestrate consensus. For one-off subtask dispatch, native Task is fine.

Synchronize state across nodes Use when native Task is wrong because the work crosses multiple agents that need to vote/sync/load-balance — TodoWrite + a single Task cannot orchestrate consensus. For one-off subtask dispatch, native Task is fine.

Manage coordination nodes Use when native Task is wrong because the work crosses multiple agents that need to vote/sync/load-balance — TodoWrite + a single Task cannot orchestrate consensus. For one-off subtask dispatch, native Task is fine.

Manage consensus protocol with BFT, Raft, or Quorum strategies Use when native Task is wrong because the work crosses multiple agents that need to vote/sync/load-balance — TodoWrite + a single Task cannot orchestrate consensus. For one-off subtask dispatch, native Task is fine.

Orchestrate multi-agent coordination Use when native Task is wrong because the work crosses multiple agents that need to vote/sync/load-balance — TodoWrite + a single Task cannot orchestrate consensus. For one-off subtask dispatch, native Task is fine.

Get coordination metrics Use when native Task is wrong because the work crosses multiple agents that need to vote/sync/load-balance — TodoWrite + a single Task cannot orchestrate consensus. For one-off subtask dispatch, native Task is fine.

Create a decentralized autonomous agent Use when native Task is wrong because you need agents that adapt their cognitive pattern (convergent / divergent / lateral / systems / critical) per-task and share knowledge across the swarm. For static one-shot agents, native Task is fine.

Trigger agent adaptation based on feedback Use when native Task is wrong because you need agents that adapt their cognitive pattern (convergent / divergent / lateral / systems / critical) per-task and share knowledge across the swarm. For static one-shot agents, native Task is fine.

Create an autonomous workflow Use when native Task is wrong because you need agents that adapt their cognitive pattern (convergent / divergent / lateral / systems / critical) per-task and share knowledge across the swarm. For static one-shot agents, native Task is fine.

Execute a DAA workflow Use when native Task is wrong because you need agents that adapt their cognitive pattern (convergent / divergent / lateral / systems / critical) per-task and share knowledge across the swarm. For static one-shot agents, native Task is fine.

Share knowledge between agents Use when native Task is wrong because you need agents that adapt their cognitive pattern (convergent / divergent / lateral / systems / critical) per-task and share knowledge across the swarm. For static one-shot agents, native Task is fine.

Get learning status for DAA agents Use when native Task is wrong because you need agents that adapt their cognitive pattern (convergent / divergent / lateral / systems / critical) per-task and share knowledge across the swarm. For static one-shot agents, native Task is fine.

Analyze or change cognitive patterns Use when native Task is wrong because you need agents that adapt their cognitive pattern (convergent / divergent / lateral / systems / critical) per-task and share knowledge across the swarm. For static one-shot agents, native Task is fine.

Get DAA performance metrics Use when native Task is wrong because you need agents that adapt their cognitive pattern (convergent / divergent / lateral / systems / critical) per-task and share knowledge across the swarm. For static one-shot agents, native Task is fine.

Initialize the ONNX embedding subsystem with hyperbolic support Use when text similarity matters beyond keyword match — native Grep finds exact strings, embeddings find meaning. Pair with memory_store / agentdb_pattern-search to land the vector against your knowledge base. For literal symbol search,…

Generate embeddings for text (Euclidean or hyperbolic) Use when text similarity matters beyond keyword match — native Grep finds exact strings, embeddings find meaning. Pair with memory_store / agentdb_pattern-search to land the vector against your knowledge base. For literal symbol search, native G…

Compare similarity between two texts Use when text similarity matters beyond keyword match — native Grep finds exact strings, embeddings find meaning. Pair with memory_store / agentdb_pattern-search to land the vector against your knowledge base. For literal symbol search, native Grep is faster.

Semantic search across stored embeddings Use when text similarity matters beyond keyword match — native Grep finds exact strings, embeddings find meaning. Pair with memory_store / agentdb_pattern-search to land the vector against your knowledge base. For literal symbol search, native Grep is faster.

Neural substrate operations (RuVector integration) Use when text similarity matters beyond keyword match — native Grep finds exact strings, embeddings find meaning. Pair with memory_store / agentdb_pattern-search to land the vector against your knowledge base. For literal symbol search, native Grep …

Hyperbolic embedding operations (Poincaré ball) Use when text similarity matters beyond keyword match — native Grep finds exact strings, embeddings find meaning. Pair with memory_store / agentdb_pattern-search to land the vector against your knowledge base. For literal symbol search, native Grep is …

Get embeddings system status and configuration Use when text similarity matters beyond keyword match — native Grep finds exact strings, embeddings find meaning. Pair with memory_store / agentdb_pattern-search to land the vector against your knowledge base. For literal symbol search, native Grep is f…

Build RaBitQ 1-bit quantized index from stored embeddings (32× compression). Pre-filters candidates via Hamming scan before exact rerank. Use when text similarity matters beyond keyword match — native Grep finds exact strings, embeddings find meaning. Pair with memory_store / agentdb_pattern-search …

Search via RaBitQ quantized index (fast Hamming scan). Returns candidate IDs for reranking. Use when text similarity matters beyond keyword match — native Grep finds exact strings, embeddings find meaning. Pair with memory_store / agentdb_pattern-search to land the vector against your knowledge base…

Get RaBitQ quantized index status — availability, vector count, compression ratio Use when text similarity matters beyond keyword match — native Grep finds exact strings, embeddings find meaning. Pair with memory_store / agentdb_pattern-search to land the vector against your knowledge base. For lite…

Analyze a GitHub repository Use when native Bash / file tools are wrong because this MCP tool exposes Ruflo-specific state or controllers that have no shell equivalent. For tasks that fit a one-line native command, prefer that.

Manage pull requests Use when native Bash / file tools are wrong because this MCP tool exposes Ruflo-specific state or controllers that have no shell equivalent. For tasks that fit a one-line native command, prefer that.

Track and manage issues Use when native Bash / file tools are wrong because this MCP tool exposes Ruflo-specific state or controllers that have no shell equivalent. For tasks that fit a one-line native command, prefer that.