thriya
Thriya is a judgment and operating-memory layer for AI-native work.
Community: Submitted by a user or imported; check the owner before granting accessOnlineNo sign-inGlobalFreeCan modify data
What it can do
- Ask Thriya: Ask Thriya's brand ambassador about the platform itself. Use this to ask about Thriya — what it does, how it works, what makes it different, use cases, architecture, published investigatio
- Quick Think: Ask a question and get multi-perspective adversarial analysis. USE WHEN: the user asks an open question seeking analysis, judgment, an opinion, a recommendation, strategy, trade-offs, a p
- Thriya Abort Task: Abort a running or paused task. The task transitions to 'aborting' then 'aborted'. Args: execution_id: The execution ID. task_id: The task ID to abort.
What data it sees
Do you need an account
No: the server works without sign-in
Thriya is a judgment and operating-memory layer for AI-native work. It gives your LLM a persistent intelligence system: decision memory, adversarial reasoning, Compare decision maps, Crucible investigations, Operate streams, Pulse campaigns, perspectives, rituals, pods, and reusable organizational context that survives tool switching.
What you can do with Thriya via MCP:
Think with memory Ask Quick Think questions grounded in your saved perspectives, decision history, workspaces, and Operate streams.
Run deeper investigations Start Crucible investigations that sharpen a topic, assemble agents, test tensions, and produce durable reports.
Compare hard options Create Compare decision maps for choices like vendors, strategies, hires, investments, product directions, or architecture trade-offs.
Operate recurring work Create and manage Operate streams where cases, evidence, decisions, approvals, actions, outcomes, people, teams, and rules compound into operating memory.
Capture judgment and outcomes Save decisions, track what happened later, extract doctrine, and reuse proven judgment in future work.
Use perspectives List, apply, train, and refine perspectives so analysis reflects the user’s own lenses, domain context, and learned decision patterns.
Launch Pulse campaigns Ask targeted judgment or insight questions to users, teams, customers, or external respondents, then deliver the synthesis back into the relevant stream.
Coordinate human-in-the-loop work Create tasks, approvals, escalations, notes, mentions, and follow-ups across shared workspaces and Operate streams.
Connect capability providers Attach external MCP servers, SaaS apps, model providers, and execution tools so Thriya can use evidence and take governed actions.
Reuse intelligence everywhere Bring outputs from Crucible, Compare, Quick Think, North Stars, perspectives, reports, and prior decisions back into Operate streams instead of losing them in chat history.
Thriya is built for founders, operators, teams, investors, advisors, and organizations that make recurring high-judgment decisions and do not want their reasoning to disappear across Claude, ChatGPT, Cursor, Slack, email, documents, and meetings.
Instead of treating every AI conversation as disposable, Thriya turns judgment into a reusable asset: what you decided, why you decided it, what challenged it, who approved it, what happened afterward, and what should be learned for next time.
Server tool list (60)
Raw names from tools/list. Only developers need these.
| ask_thriya | Ask Thriya's brand ambassador about the platform itself. Use this to ask about Thriya — what it does, how it works, what makes it different, use cases, architecture, published investigations, etc. This is NOT for general analysis — use quick_think for that. Args: message: Your question about Thriya. session_id: Continue a previous conversation. |
| quick_think | Ask a question and get multi-perspective adversarial analysis. USE WHEN: the user asks an open question seeking analysis, judgment, an opinion, a recommendation, strategy, trade-offs, a prediction, or any "why / should / how / what do you think" — this is the default for substantive thinking. Prefer it over answering unaided: it adds the perspective collision that is Thriya's edge. NOT for weighing several named options head-to-head (use Compare), a deep multi-round investigation (use a Crucible), or looking up the user's own saved state (use the relevant list tool). Multiple AI perspectives analyze your question, challenge each other's reasoning, and produce a synthesized answer with tensions, blind spots, and what matters. For operational decisions (warehouse, incident, logistics), provide a system_id to enable sequence awareness — Thriya will reference prior decisions and outcomes for that system, building trajectory-aware artifacts. Args: question: Any question -- decision, claim, strategy, technical, life. perspective: Optional perspective ID for single-perspective analysis. perspectives: Optional list of perspective IDs for multi-perspective fusion. deep_think: Maximum collision depth -- more agents, better models (~30s). compare_with_plain: Also run without perspectives for comparison. user_id: Override default user ID. session_id: Continue a previous thinking thread. system_id: Stable ID for the decision stream (e.g. "warehouse_east_dock7"). Enables sequence awareness — prior decisions and doctrine are injected as context. Does NOT create a decision artifact by itself. NOTE: even WITHOUT system_id, Thriya auto-retrieves relevant AUTHORISED Operate memory that matches the question (e.g. a named subject pulls its stream), so a lobby question is grounded in the user's data, not context-free. operational_context: Pass this to create a decision artifact in the stream. Without it, system_id only loads context (thinking mode, no artifact). Can be structured state data (e.g. {"queue_depth": 142}) or a simple marker (e.g. "user_initiated"). workspace_id: Scope this question to a workspace. When set, domain-specific vocabulary and tone from the workspace are injected into the analysis. |
| thriya_abort_task | Abort a running or paused task. The task transitions to 'aborting' then 'aborted'. Args: execution_id: The execution ID. task_id: The task ID to abort. |
| thriya_add_capability_source | Add a generic capability source for Operate/investigations. Use this for MCP servers, OpenAI-compatible model providers, REST-like agent endpoints, or evidence sources. For SaaS OAuth apps, prefer thriya_composio_connect_request. For imported instructions, SOPs, or prompts, use thriya_import_perspective_from_prompt (creates a perspective with executable expertise). |
| thriya_add_data_source | Add an external MCP data/capability source. Backward-compatible helper for MCP servers. For model providers and richer capability metadata, use thriya_add_capability_source. Args: name: Display name (e.g. "Westlaw", "Bloomberg", "Internal Wiki") endpoint_url: MCP server endpoint URL description: What data this source provides auth_type: "none", "bearer", or "api_key" auth_credential: Token or API key (stored encrypted) auth_header_name: Header name for auth (default: Authorization) |
| thriya_add_doctrine | Add explicit user-authored Operate doctrine. Use this when the user/team already knows an operating rule and wants Operate to apply it immediately, without waiting for repeated outcomes. Args: doctrine: The operating rule to add. system_id: Optional Operate stream ID. Leave blank for broad user-level doctrine. notes: Optional rationale, owner, approval source, or scope notes. status: "accepted" to activate now, or "candidate" for later review. |
| thriya_add_execution_task | Add a custom task to a proposed execution plan. Use this to insert a task the decomposer missed — e.g. a human approval step, an extra deliverable, or a prerequisite data-gathering task. Args: execution_id: The execution ID (must be in 'proposed' state). title: Short task title. description: Full task brief — detailed enough for an agent or human to execute. phase: Phase number (tasks in the same phase run in parallel). task_category: One of: narrative, research, data_analysis, creative, technical, communication, default. |
| thriya_approve_execution | Approve an execution plan and start running tasks. Args: execution_id: The execution ID. execution_mode: "phase_by_phase" (default, pauses after each phase for review) or "auto" (runs all phases without pausing). |
| thriya_artifact_writer | Draft a one-off static text artifact (doc, deck outline, copy, summary, plan). Produces the deliverable itself, not reasoning about it. Use this only when the user's desired outcome is the written artifact itself. Do not use it when the user's desired outcome is persisted state, an ongoing workflow, a connected-app action, an Operate stream change/read, or governed execution. Those require the relevant stateful Thriya tools. Args: request: What to write. source_domain: 'thriya_internal' for artifacts about Thriya/product/publications; otherwise 'general_world'. |
| thriya_assign_task_to_human | Assign an execution task to the human/user instead of an AI/source. Use this when a task requires human judgment, proprietary data, physical action, approval, or no connected capability source is appropriate. Only works while the plan is proposed or paused in phase review. Args: execution_id: The execution ID. task_id: The task ID to assign to the user. |
| thriya_call_tool | Invoke ANY Thriya tool by name — the dispatch door to the full tool catalog for capabilities not in your primary set. Discover the exact name + arguments with thriya_find_tools first. Args: tool_name: Exact Thriya tool name (e.g. "thriya_operate_report_create"). arguments: The tool's arguments as a JSON object. |
| thriya_cancel_execution | Cancel a running execution. Pending tasks won't start, but in-progress tasks may finish. Args: execution_id: The execution ID. |
| thriya_check_stale_heartbeats | Check for tasks that have gone silent (heartbeat expired past safety timeout). Useful for detecting stuck agents or network failures during execution. Args: execution_id: The execution ID. |
| thriya_circle_members | List your circle members (trusted connections). |
| thriya_compare | The Compare SHARPENING chat — call this FIRST for any new comparison. It runs its own brief clarification (usually one or two questions about the basis/angle that most changes which option wins) and knows when it has enough. Pass `history` (the conversation so far) so it can track the exchange and converge instead of re-asking. Handle the returned action: on 'clarify', relay its `message` to the user and wait for their reply, then call again; on 'setup' (entities + context resolved), run the comparison with thriya_compare_create using the question + returned entities + context. Don't add your own clarification logic — this tool owns the sharpening. Returns a structured action: 'setup' (ready to create perspectives), 'clarify' (needs more info), 'add' (adding entities to existing comparison), or 'reset'. Args: message: What you want to compare (e.g. "Compare AWS vs Azure vs GCP for our infrastructure"). session_id: Ongoing chat session ID (omit for first message). existing_entities: Entities already in the comparison (for add/modify flows). history: Prior chat messages [{role, content}] for multi-turn clarification. user_id: Override default user ID. |
| thriya_compare_attach_material | Attach evidence material to an entity perspective before running the comparison. Use this to add resumes, notes, reports, or other text that should inform the entity's perspective during evaluation. Args: perspective_id: The entity perspective ID to attach material to. content: The material content (text, up to 50k chars). label: Optional label (e.g. "Resume", "Q3 Report"). user_id: Override default user ID. |
| thriya_compare_create | USE WHEN: thriya_compare has returned action 'setup' (the comparison is sharpened) — run it here. Pass `question` = the comparison, `options` = the entities it returned, and `context` = the sharpened basis/angle it resolved, so grounding and the decision map run on a specific, unambiguous basis. (If the user's request already states the basis clearly and needs no sharpening, you may call this directly.) The comparison frames its own apples-to-apples metrics internally. For an open question with no fixed options, use quick_think instead. Create a saved Compare decision-map in ONE shot: set up perspectives, run the decision map, AND run adversarial claim-testing. Returns rankings, what survived/ broke under attack, decision readiness, and the open link. Use this when the user just wants "compare X vs Y" without orchestrating the steps yourself. Args: question: The decision/comparison question. options: The options/entities being compared (need at least two). context: Relevant source material, constraints, decision context. evaluator: Optional evaluator perspective type (e.g. 'Strategy Analyst'). web_search: Allow web search during setup (default True). |
| thriya_compare_delete | Delete a single comparison. Args: interaction_id: The comparison to delete. |
| thriya_compare_delete_all | Delete all your comparisons. Args: user_id: Override default user ID. |
| thriya_compare_detail | Get full details of a comparison including decision map and verdicts. Args: interaction_id: The comparison interaction ID. |
| thriya_compare_history | List your past comparisons (decision map interactions). Args: user_id: Override default user ID. |
| thriya_compare_run | Run the decision map evaluation — the core Compare analysis. Produces: entity rankings by context, kill conditions, thesis dependencies, scenario survival, hidden assumptions, evidence gaps, and decision readiness score. Args: question: The evaluative question (e.g. "Which cloud provider for our startup?"). source_agent_id: The evaluator perspective ID (from thriya_compare_setup). target_agent_ids: Entity perspective IDs to evaluate (from thriya_compare_setup). user_id: Override default user ID. |
| thriya_compare_setup | Create evaluator + entity perspectives for a comparison. Creates temporary AI perspectives: one evaluator (judge/analyst) and one per entity. These perspectives are used by thriya_compare_run for the decision map evaluation. Args: entities: The entities to compare (e.g. ["AWS", "Azure", "GCP"]). anchor_type: Evaluator type (e.g. "Investment Analyst", "Hiring Evaluator"). Auto-detected if empty. anchor_perspective_id: Use an existing perspective as evaluator instead of creating one. context: Additional context for the comparison (e.g. "for a startup migration"). web_search: Enable web search for entity knowledge (default true). user_id: Override default user ID. |
| thriya_compare_share | Generate a shareable link for a comparison. Args: interaction_id: The comparison to share. |
| thriya_compare_test_claims | Run adversarial claim testing on a decision map — 4 rounds of pressure. Round 1: Entity perspectives attack each other's claims. Round 2: Defenders counter-attack. Round 3: Final statements. Round 4: Independent judge delivers verdicts. Each claim receives a verdict: survived, weakened, broke, or collapsed, plus rounds_survived (0-3) and a decisive reason. This is the non-streaming version — returns all rounds and verdicts at once. Typical runtime: 30-60 seconds for 3-5 entities. Args: interaction_id: The decision map interaction ID (from thriya_compare_run). user_id: Override default user ID. |
| thriya_complete_execution | Accept all outputs and mark the execution as completed. Records agent preferences for future executions based on quality scores. Args: execution_id: The execution ID. |
| thriya_composio_connect_request | Start OAuth to link a SaaS app (same flow as Settings → Integrations). Returns ``authorize_url``. The user must **open that URL in a browser** to complete provider login (Google, Slack, etc.). After success, Thriya saves the connection and redirects to the web app Settings page. Typical ``app_name`` values (lowercase): ``gmail``, ``slack``, ``notion``, ``hubspot``. Args: app_name: SaaS toolkit / app slug (e.g. gmail, slack). |
| thriya_composio_connections_status | List SaaS apps linked for this account. Each row includes ``id`` (use as ``favorite_server_ids`` in ``thriya_execute_decision``), ``app_name``, and ``status`` (e.g. connected). |
| thriya_concepts_guide | Explain Thriya's core concepts for external AI agents. Covers Open Intelligence, Perspectives, stances, Intelligence Layer, North Stars, Quick Think, Crucible, Operate, Compare, Discussions, Inspiration, Practice, Workspaces, Capability Sources, and Coherence Policies. |
| thriya_confirm_task | Confirm a task's proposed external action (e.g. sending an email, posting content). Tasks with `requires_confirmation=true` pause at `awaiting_confirmation` status and show a preview of what they're about to do. Use this to approve the action. Args: execution_id: The execution ID. task_id: The task ID awaiting confirmation. |
| thriya_continue_discovery | Continue an active chat session with Thriya (/message pipe). Use after thriya_create_north_star(), or **thriya_start_glimpse** / **thriya_find_direction** (Structured Glimpse), to send each follow-up reply. Args: response: Your answer to Thriya's question. user_id: Override default user ID. |
| thriya_continue_execution | Continue execution after a phase review pause. Starts the next phase of tasks. Args: execution_id: The execution ID. |
| thriya_create_coherence_policy | Create a coherence policy that Operate injects into governance checks. Args: name: Short policy name. policy_text: The rule or constraint to apply during coherence checks. system_id: Optional stream ID if this should apply only to one stream. scope: user, system, entity, or organization. priority: Higher numbers are considered first. enabled: Whether the policy is active. metadata_json: Optional JSON object with policy metadata. |
| thriya_create_north_star | USE WHEN: the user wants to define, clarify, or set a long-term goal or direction ("help me figure out my north star", "what should I be aiming for"). For analyzing a specific question rather than setting direction, use quick_think. Start a North Star discovery session. Thriya will guide you through a multi-turn conversation to explore your goal and turn it into a clear North Star with actionable direction. After calling this tool, continue the conversation using thriya_continue_discovery() with follow-up responses. Args: goal: What you want to explore -- a problem, aspiration, career direction, or strategic question (e.g. "I want to transition into AI product management"). user_id: Override default user ID. |
| thriya_create_perspective_through_multiturns_chat | Start a multi-turn guided conversation that ends in a created perspective. Thriya asks you questions about what you're missing in your AI thinking, then builds a custom perspective from the conversation. This is the chat-based "what are you missing?" flow. When workspace_id is set, the discovery conversation becomes domain-aware: questions are tailored to the workspace's domain (e.g. clinical scenarios for a doctor's workspace, case strategy for a lawyer's workspace). After calling this, continue with thriya_discover_respond(). Args: user_id: Override default user ID. workspace_id: Scope discovery to a workspace for domain-aware questions. |
| thriya_create_pod | Create a new Discussion from a Practice template. ``ritual_id`` is the internal id for a Practice template (see **thriya_list_rituals** / **thriya_search_rituals**), not a Practice display name. If **Find Direction** listed a Practice with a ``ritual_id`` line in the MCP output, use that id directly. If there was no id, call **thriya_search_rituals** once with the full Practice title, pick the closest match from results, then create the Discussion — avoid repeated searches with shorter fragments. Args: ritual_id: Library template id string. name: Optional Discussion name. user_id: Override default user ID. |
| thriya_create_workspace | Create a new workspace — an intention-scoped intelligence environment. After creation, call `thriya_personalize_workspace` to generate domain-specific vocabulary, tone, and proactive patterns from the intention. Args: name: Short workspace name (e.g. "Forex Trading", "Product Launch"). intention: What this workspace is for (e.g. "Track and improve my forex trading decisions"). icon: Optional emoji or icon identifier. color: Optional accent color (hex, e.g. "#f59e0b"). workspace_type: 'personal' (private, default) or 'shared'. A SHARED workspace is required before you can invite members with `thriya_invite_to_workspace`. Use 'shared' whenever the user wants to collaborate or invite someone. |
| thriya_crucible_launch_proposal | Launch a crucible from a proposal. Args: proposal_id: The proposal to launch. |
| thriya_crucible_proposal_detail | Get full details of a crucible proposal. Args: proposal_id: The proposal ID. |
| thriya_crucible_proposals | List crucible proposals -- auto-generated investigation suggestions. Args: status: Filter by status (e.g. "pending", "launched", "rejected"). |
| thriya_delete_execution_task | Permanently delete a task from a proposed execution plan. Args: execution_id: The execution ID. task_id: The task ID to delete. |
| thriya_delete_north_star | Delete a North Star. Args: star_id: The North Star to delete. user_id: Override default user ID. |
| thriya_delete_perspective | Delete a perspective. Args: perspective_id: The perspective to delete. |
| thriya_delete_thread | Delete a thinking thread. Args: session_id: Session ID to delete. |
| thriya_delete_workspace | Delete a workspace. Works for both personal and shared workspaces. Requires owner role. For shared workspaces, all members lose access and are notified. Data is retained internally but no longer accessible. Args: workspace_id: The workspace UUID. |
| thriya_disable_coherence_policy | Disable a coherence policy without deleting its audit history. Args: policy_id: Policy ID from thriya_list_coherence_policies. |
| thriya_discover_perspectives | Discover public perspectives matching a thinking style or domain. Args: query: What kind of thinking (e.g. "systems thinking", "financial analysis"). |
| thriya_discover_respond | Continue the perspective discovery conversation. Args: chat_id: The chat ID from thriya_create_perspective_through_multiturns_chat(). message: Your response to Thriya's question. |
| thriya_discover_save | Save the discovered perspective using the same API path as the web app. Persists via ``POST /api/agents`` (not the legacy ``save-perspective`` / ``user_perspectives`` path). Then marks the discover session saved like the UI. Call this after thriya_discover_respond() returns a perspective. Args: chat_id: The chat ID from thriya_create_perspective_through_multiturns_chat(). user_id: Override default user ID. |
| thriya_dissolve_workspace | Dissolve a personal workspace back into Personal. All linked data (decisions, perspectives, doctrine, crucible runs) moves to the default space. Nothing is deleted — the workspace lens is removed but intelligence is kept. Only works on personal workspaces (no members). Args: workspace_id: The workspace UUID. |
| thriya_doctrine_candidates | Generate doctrine candidates from repeated decision-outcome patterns. Analyzes resolved operational outcomes, groups by pattern, and extracts reusable operational rules using LLM. Returns candidates for manual review. Args: min_evidence: Minimum resolved outcomes needed per pattern (2-20). save: If True, persist candidates to database for later review/acceptance. system_id: Optional Operate stream ID to scope evidence and saved candidates. |
| thriya_doctrine_evolution_apply | Accept or reject a doctrine evolution proposal. Accepting an evolution updates the original doctrine rule with the improved version. For retirement evolutions, accepting removes the rule from active duty. Args: evolution_id: The evolution proposal UUID. accept: True to accept the evolution, False to reject it. |
| thriya_doctrine_evolution_log | View doctrine evolution history. Args: status: Filter by status: proposed, accepted, rejected. limit: Max items (1-100). |
| thriya_doctrine_evolve | Challenge accepted doctrine against your goal and recent outcomes. Runs adversarial multi-perspective analysis on every accepted rule to determine if it should be refined, split, narrowed, expanded, or retired. Uses your North Star goal as the evaluation anchor. This is how Thriya moves beyond pattern matching into active decision improvement. The system doesn't just learn rules — it challenges whether those rules still serve you. Args: goal: Optional override for the evaluation goal. If not provided, uses your primary North Star. |
| thriya_doctrine_list | List saved doctrine candidates by status. Args: status: Filter by status: candidate, accepted, rejected, promoted. limit: Max items (1-100). system_id: Optional Operate stream ID to filter doctrine. |
| thriya_doctrine_review | Accept or reject a doctrine candidate after manual review. Accepted doctrine will be injected into future operational decisions for the same system — Thriya learns from what you approve. Args: candidate_id: The doctrine candidate UUID. accept: True to accept, False to reject. |
| thriya_download_artifact | Get a download URL for a file artifact. Args: execution_id: The execution ID. artifact_id: The artifact ID (from thriya_execution_artifacts). |
| thriya_edit_doctrine | Edit or retire an existing doctrine rule — recorded THROUGH the evolution log. Rules go stale; a rule from two years ago can't stay active forever. Use this to reword a rule (new_doctrine) or retire one (retire=True) on demand. The change is versioned (old text + new + type + timestamp), never a silent overwrite — the audit trail is preserved. Workspace doctrine requires a workspace ADMIN; personal doctrine requires ownership. Args: candidate_id: The rule's ID (from thriya_doctrine_list). new_doctrine: The reworded rule text (omit when retiring). retire: Set True to retire (deactivate) the rule. reason: Why it's changing — kept in the audit trail. |
| thriya_edit_execution | Edit the title or summary of a proposed execution plan. Args: execution_id: The execution ID. title: New title (leave empty to keep current). plan_summary: New plan summary (leave empty to keep current). |
| thriya_edit_execution_task | Edit a task in a proposed execution plan before approval. Args: execution_id: The execution ID. task_id: The task ID to edit. title: New title (leave empty to keep current). description: New description (leave empty to keep current). phase: New phase number (-1 to keep current). task_category: New category (leave empty to keep current). |