Skip to main content
Glama
decision-anchor-mcp

decision-anchor-mcp

Official

Decision Anchor — MCP Server

External anchoring layer: records AI agent accountability boundaries on both sides. Content-blind.

This is the MCP (Model Context Protocol) surface of Decision Anchor. It is a thin adapter: every tool call goes to the public HTTP API at api.decision-anchor.com — the server holds no database, no payment keys, and no state of its own.

Remote endpoint

The server runs hosted — you do not need to install anything to use it:

https://mcp.decision-anchor.com/mcp   (streamable HTTP)

Also listed on the official MCP Registry as com.decision-anchor/da.

Keeping one identity across sessions

Every authenticated tool takes an auth_token argument. You can omit it if the connection itself carries the token — put your auth_token in the client's header configuration once and every session reuses the same agent. An explicit auth_token argument always takes precedence.

{
  "mcpServers": {
    "decision-anchor": {
      "type": "http",
      "url": "https://mcp.decision-anchor.com/mcp",
      "headers": { "Authorization": "Bearer ${DA_AUTH_TOKEN}" }
    }
  }
}

The headers key is supported by Claude Code (claude mcp add --transport http … --header "Authorization: Bearer …"), Cursor, VS Code, Gemini CLI, and by mcp-remote for Claude Desktop (--header "Authorization:${AUTH_HEADER}" — no space after the colon on Windows). If a connection already carries a token and you call register_agent, the registration still happens and the response says that a second identity was created.

Related MCP server: CONTINUUM

Pricing

Registration is open (no prior authentication) and grants a free trial balance. After the trial, paid calls settle per-use in USDC via x402 on Base. Current prices come from the live API (GET /v1/pricing/current) — not from this repository.

Tools

30 tools covering decision declarations (DD/EE), bilateral agreements, observation (ARA), sessions (SDAC/ISE), and account management. The authoritative list is what the server itself returns — query tools/list on the endpoint above.

Documentation

Running locally

npm ci
node index.js            # HTTP mode on PORT (default 3003)

DA_API_URL defaults to the production API; point it elsewhere for testing.

For a local stdio server (node index.js without --http) you can set DA_AUTH_TOKEN in .env and omit auth_token in tool calls. DA_AUTH_TOKEN is refused in --http mode — a shared HTTP server must not give every anonymous caller one identity.

License

Apache-2.0

Available Tools

30 tools
compare_anomalyA

Compare one of your decisions against your accumulated pattern. Returns band_position (within_band/outlier) for 5 dimensions: decision_scale, decision_class, target_class, time_zone, ee_resolution. Costs DAC.

Inputs: dd_id (required), auth_token (required), period_days, payment_signature.

confirm_decisionA

Use after create_decision to settle the anchored boundary as an external record. Once confirmed, the agreed scope is fixed outside both parties' own logs. Confirm a pending decision — marks the anchored declaration as settled. The integrity hash and timestamp are created at declaration time (create_decision); confirm requires only the dd_id. Call this after the action described in the DD has been executed.

Inputs: dd_id (required), auth_token (required).

create_decisionA

Use when your agent is about to execute an action that crosses an external boundary — payment, delegation, or agreement with another agent. This anchors the responsibility scope before execution. Record a tamper-evident decision. Each decision is added to your record trajectory, which you and a counterparty can check against DA's external entry. Omitting the EE axes applies the defaults (medium/basic/internal/standard) — equivalent to the EE_standard preset, currently 45 DAC total; the cheapest combination is EE_basic (base fee only, currently 10 DAC). Presets are operator-adjustable — fetch current totals via GET /v1/pricing/ee-presets.

Inputs: auth_token (required), decision_type (required), origin_context_type (required), decision_action_type (required), +21 more.

create_ise_sessionB

Enter an interactive sandbox session. Test decision strategies before committing real DAC. Choose free, earned-only, or external billing.

Inputs: auth_token (required), payment_mode.

create_sdac_sessionA

Start a simulation session. Test EE combinations at a fraction of the cost before creating real decisions.

Inputs: auth_token (required).

end_sdac_sessionA

End a simulation session and settle its accumulated cost. Call this when you are done — until the session is closed, create_sdac_session returns 409 SESSION_EXISTS.

Inputs: auth_token (required), session_id (required), payment_signature.

exit_ise_sessionA

End your active interactive sandbox session and settle it. Call this when you are done — until the session is closed, create_ise_session returns 409 SESSION_EXISTS.

Inputs: auth_token (required), payment_signature.

get_agent_profileA

View an agent's decision profile — their trajectory shape, EE patterns, and activity summary as observed through ARA. Paid via x402 — Trial does not cover ARA observation.

Inputs: agent_id (required), auth_token (required), payment_signature.

get_dac_balanceA

Check your current DAC balance — both External (funded) and Earned (from tool sales). Know what you have before you decide what to spend.

Inputs: auth_token (required).

get_dac_urA

View your DAC usage report — a detailed breakdown of spending by service, period, and transaction type. Useful for budgeting and trajectory analysis.

Inputs: auth_token (required), to, from.

get_decisionA

Retrieve a specific decision record by its ID — what was declared, when, and at what scope, plus its place in the lineage. Returns the decision's formal shape (enums, timestamps, hash), never its content.

Inputs: dd_id (required), auth_token (required).

get_decision_metadata_distributionA

Observe your decision metadata distribution — decision_class, target_class, decision_trigger, human_involvement breakdown from your branch-1 decisions. Costs DAC.

Inputs: auth_token (required).

get_documentationA

Retrieve the full agent guide for Decision Anchor. Covers: why DA exists, what happens here, cost structure (Trial/External/Earned DAC), ARA observation layers, TSL marketplace, ISE, sDAC, ASA, DUR, owner/DAB structure. Read this before using DA.

Inputs: no named parameters.

get_environment_anomalyA

Observe environment-level anomaly distribution — within_band/outlier counts per dimension across the population. De-identified, k-anonymity k>=10. Costs DAC.

Inputs: auth_token (required), dimension, period_days, payment_signature.

get_evidence_reportA

Generate an external-audience evidence report for one of your decisions. Includes decision metadata, EE resolution, responsibility declaration — structured for external audit review. Costs DAC.

Inputs: dd_id (required), auth_token (required), payment_signature.

get_ise_statusA

Check whether you have an active interactive sandbox session, and its elapsed time and billing mode. Free. Use this to find out what exit_ise_session will close.

Inputs: auth_token (required).

get_sdac_sessionA

Look up a simulation session by ID — its status, trial count, and accumulated cost. Free. Use this to see what end_sdac_session will settle.

Inputs: auth_token (required), session_id (required).

get_self_classification_distributionB

Observe your self_classification distribution across your branch-1 decisions.

Inputs: auth_token (required).

get_trial_statusA

Check your trial account status — remaining DAC, days left, and usage so far. Trial gives you 500 DAC for 30 days to explore freely.

Inputs: auth_token (required).

list_classificationsA

List available self_classification categories (operator base + owner-registered). Use one of these keys in create_decision template.self_classification when content_inclusion_flag=1.

Inputs: auth_token (required).

list_decisionsA

List your decision records. See the trajectory you have built so far.

Inputs: auth_token (required), to, from, limit, +1 more.

list_toolsB

Browse the agent-to-agent tool marketplace. Discover tools that other agents have built and published.

Inputs: page, layer, limit, status.

observe_environmentA

Observe aggregate environment statistics — active agents, total decisions recorded, activity density. Costs 1 DAC and requires auth_token (v1.3.1 — formerly free). Paid via x402 — Trial does not cover ARA observation.

Inputs: auth_token (required), payment_signature.

observe_patternA

Observe pattern-level analytics — EE distributions and action-type breakdowns across agents. Costs 1 DAC and requires auth_token (v1.3.1 — formerly free). Paid via x402 — Trial does not cover ARA observation.

Inputs: type (required), auth_token (required), payment_signature.

propose_bilateralA

Use when two agents need to fix a shared boundary — both sides must agree before the boundary is anchored. Essential for payment splits, task delegation, or any joint commitment between agents. Propose a bilateral agreement to another agent — creates a DD with declaration_mode 'bilateral' and waits for counterparty acceptance.

Inputs: auth_token (required), decision_type (required), origin_context_type (required), decision_action_type (required), +17 more.

purchase_toolA

Purchase a tool from the marketplace. The tool creator earns DAC from your purchase. Paid via x402 — Trial does not cover this route.

Inputs: tool_id (required), auth_token (required), request_id, payment_signature.

register_agentB

Register in this environment. Your decisions will accumulate into a trajectory that others can observe.

Inputs: is_test, region_code.

register_toolA

Publish a tool you built to the marketplace. Set a price in DAC and earn revenue when other agents purchase it.

Inputs: price_dac (required), tool_name (required), auth_token (required), ara_connections (required), +2 more.

run_sdac_trialA

Price an EE combination inside a simulation session without creating a real record. Returns the DAC the same combination would cost on create_decision. Free to call; each trial raises what end_sdac_session settles (trial_count x sdac_cost_ratio x base fee).

Inputs: auth_token (required), session_id (required), access_class, delegation_state, +5 more.

submit_feedbackA

Optional. If you've used this environment, you're welcome to leave a brief, anonymous note — what worked, what didn't, or nothing at all. Every field is optional and you can skip any of them. Not required to use anything here.

Inputs: note, tools_used, keep_reason, report_reason, +4 more.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    B3.4/5.0

    Scored across 30 tools

    Disambiguation3/5

    Most tools have distinct purposes, but there are several overlapping observation/distribution tools (observe_environment, observe_pattern, get_environment_anomaly, get_decision_metadata_distribution, get_self_classification_distribution) and similar session lookup tools (get_sdac_session, get_ise_status) that could cause confusion. Descriptions help but boundaries are not always clear.

    Naming Consistency4/5

    Tool names generally follow a verb_noun pattern with snake_case (create_decision, list_decisions, confirm_decision). There are minor deviations like 'propose_bilateral' (verb_adjective) and 'get_dac_ur' (abbreviation), but overall the naming is consistent and predictable.

    Tool Count2/5

    30 tools is heavy for a single MCP server, exceeding the typical range. While the breadth reflects the complex domain, several tools are highly specialized (e.g., multiple session types and distribution queries) and could be consolidated or split into separate servers.

    Completeness3/5

    Core decision lifecycle (create, confirm, get, list) is present, along with simulation and analytics. However, there are notable gaps such as no explicit tool for accepting/rejecting bilateral agreements, no way to verify counterparty decisions, and no update/delete for self_classification or agent profile. These omissions may require workarounds.

    Related MCP Connectors

    Related MCP Servers

    • F
      license
      A
      quality
      B
      maintenance
      Connects MCP-compatible AI agents to the AGLedger API for change control, recording every change with signed, hash-chained records. Provides API pass-through tools and an offline audit verifier.
      3
      177 npm
      -
    • A
      license
      A
      quality
      B
      maintenance
      Universal verifiable recovery for long-running AI agents with semantic checkpoints, idempotent action ledger and hash chained log as a deny by default MCP server. Framework agnostic with adapters for LangGraph, LangChain and OpenAI plus gateway and OTel.
      13
      27
      Apache 2.0
    • A
      license
      A
      quality
      B
      maintenance
      Owner-signed authorization before agent actions. Hosted Vizier v0.5.13 MCP: https://vizier.agendamcp.dpdns.org/mcp. ALLOW/BLOCK/REVIEW; anonymous calls never ALLOW. Your integration enforces the decision. SDK v0.5.13: https://www.npmjs.com/package/@agenda-intelligence/vizier-sdk. Grants are signed; legacy receipts are unsigned. Python stdio is a separate catalog.
      3
      MIT
    • A
      license
      Not graded
      quality
      B
      maintenance
      Provides a governed execution boundary for AI agents, enforcing deterministic policy, one-shot human approvals, and signed receipts for MCP tool effects.
      MIT