Agent Provider Depot MCP

Session start, if your code calls third-party APIs: what changed for your stack since your last call (retired models, price changes, deprecated packages, pick changes). Send provider, model and package ids (never a manifest) and your last cursor. Returns clear|changed|attention, at most 3 actions (source, falsifier), a new cursor. New choice: decide. Live error: rescue.

{
  "type": "object",
  "properties": {
    "cursor": {
      "type": "string",
      "description": "From your last brief"
    },
    "models": {
      "type": "array",
      "items": {
        "type": "string"
      },
      "description": "Model ids you call"
    },
    "packages": {
      "type": "array",
      "items": {
        "type": "string"
      },
      "description": "id or id@version"
    },
    "policies": {
      "type": "object",
      "properties": {
        "maxMonthlyUsd": {
          "type": "number",
          "minimum": 0
        },
        "requireVerifiedOutcome": {
          "type": "boolean"
        }
      }
    },
    "providers": {
      "type": "array",
      "items": {
        "type": "string"
      },
      "maxItems": 50,
      "description": "e.g. anthropic, neon"
    },
    "stackFingerprint": {
      "type": "string",
      "description": "sha256:<64 hex>, hashed locally"
    }
  }
}

{
  "title": "Every-session brief",
  "readOnlyHint": true,
  "openWorldHint": false,
  "idempotentHint": true,
  "destructiveHint": false
}

Diagnose a failing third-party API call before you retry or swap: paste the exact error (429, 402, 529 overloaded, insufficient_quota, model_not_found, deprecated SDK) and provider id. Returns the cause, an immediate fix and a fallback route, or no match plus checks to run. Not for picking a provider (decide).

{
  "type": "object",
  "properties": {
    "q": {
      "type": "string",
      "description": "Error text, status or SKU"
    },
    "sku": {
      "type": "string",
      "description": "Model or SKU id"
    },
    "jtbd": {
      "type": "string",
      "description": "Job id"
    },
    "status": {
      "type": "string",
      "description": "HTTP status, e.g. 429"
    },
    "provider": {
      "type": "string",
      "description": "Provider id that failed"
    }
  }
}

{
  "title": "Rescue a failing API call",
  "readOnlyHint": true,
  "openWorldHint": false,
  "idempotentHint": true,
  "destructiveHint": false
}

Choose a provider before writing integration code: LLM API, code sandbox, vector DB, web search, email, Postgres, auth, browser, observability. Send task in words or a jtbd id. Returns the dated editorial Depot Pick, a runner-up, the pick's price with source URL and date, when not to take it, and reportOutcome (for report_outcome). Quotes, comparisons, audits: depot.

{
  "type": "object",
  "properties": {
    "full": {
      "type": "boolean"
    },
    "jtbd": {
      "type": "string",
      "description": "Or a job id, e.g. agent-sandbox; wins over task"
    },
    "task": {
      "type": "string",
      "description": "What you need, in words"
    },
    "existing": {
      "type": "string",
      "description": "Comma list of provider ids you already use"
    },
    "maxMonthly": {
      "type": "number",
      "minimum": 0,
      "description": "USD budget for autoActOk"
    },
    "constraints": {
      "type": "string",
      "description": "Comma list, e.g. eu-region; may switch the pick"
    }
  }
}

{
  "title": "Pick a provider",
  "readOnlyHint": true,
  "openWorldHint": false,
  "idempotentHint": true,
  "destructiveHint": false
}

One op per call, args as an object: report_outcome(decision, provider, worked) after acting; quote(jtbd) sourced SKU prices; estimate(jtbd, workload); compare(providers); free(jtbd); check(providers) deprecated/retired flags; audit_repo(packages, models); stack(task), bundle(jtbds); search(q); kit(class); answer(task); stack_delta(providers, since); changes(since); failures; coverage; freshness. op=help lists args.

{
  "type": "object",
  "required": [
    "op"
  ],
  "properties": {
    "op": {
      "enum": [
        "answer",
        "estimate",
        "search",
        "stack",
        "quote",
        "bundle",
        "kit",
        "compare",
        "free",
        "failures",
        "coverage",
        "changes",
        "check",
        "freshness",
        "audit_repo",
        "stack_delta",
        "report_outcome",
        "help"
      ],
      "type": "string"
    },
    "args": {
      "type": "object"
    }
  }
}

{
  "title": "Depot drill-down",
  "readOnlyHint": false,
  "openWorldHint": false,
  "idempotentHint": true,
  "destructiveHint": false
}

Advertised by the server during the last successful anonymous check.

Every-session check for a declared stack: what changed since your last cursor, at most 3 actions, and a new cursor.

providers

Comma-separated provider ids you use (ids only, never a manifest) (required)

Advertised by the server during the last successful anonymous check.

Resolve a multi-job agent architecture or JTBD into verified Depot Picks, runner-up escape hatches, and sourced SKU quotes (<3 KB JSON).

task

The agent job or stack you are choosing infrastructure for (e.g. coding-agent, voice-agent, web-scraping) (required)

Advertised by the server during the last successful anonymous check.

Diagnose a runtime API failure (HTTP 429, 402, 529, 413, Cloudflare 1015/1020, model_not_found) or deprecated SDK and get an immediate code fix + hot-swap fallback route.

error

Exact error string, HTTP status code, or deprecated model/package name (required)

Advertised by the server during the last successful anonymous check.

Generate a 4-stage Agent Evaluation & Task Benchmark plan (public benchmark, sandbox harness, CI pass^3 regression gate, and LLM-as-judge calibration).

task

Agent domain to evaluate (coding, customer-support, rag, browser, computer-use, voice, security) (required)

Advertised by the server during the last successful anonymous check.

Audit a repo dependency list or package.json for retired model SKUs, wrong npm packages, and cheaper verified alternatives (<2 KB JSON).

deps

Comma-separated npm/pip packages or provider slugs (e.g. openai,langchain,pinecone-client,resend) (required)

Advertised by the server during the last successful anonymous check.

Pick providers via JTBD depot picks; cite Depot Pick + methodology

task

The agent job or stack you are choosing infrastructure for (required)

Advertised by the server during the last successful anonymous check.

Check an existing provider stack for deprecations, stale prices, and compliance gates (soc2, dpa, zdr)

providers

Comma-separated provider slugs (required)

Advertised by the server during the last successful anonymous check.

Inspect deterministic falsification thresholds and runner-up switch triggers for a JTBD pick

jtbd

JTBD slug (e.g. coding-agent, managed-vector) (required)

Advertised by the server during the last successful anonymous check.

Run typed Agent Suite v2 with traces for official classification

agentType

Agent class (e.g. coding, rag, browser)

Advertised by the server during the last successful anonymous check.

What Depot is, entrypoints, counts

https://getagentdepot.com/api/hub

MIME type: application/json

Advertised by the server during the last successful anonymous check.

Definitive agent map (< 5 KB)

https://getagentdepot.com/llms.txt

MIME type: text/plain

Advertised by the server during the last successful anonymous check.

If you are an agent, start here

https://getagentdepot.com/agents.md

MIME type: text/markdown

Advertised by the server during the last successful anonymous check.

Depot Hub skill instructions (SKILL.md)

https://getagentdepot.com/skill.md

MIME type: text/markdown

Advertised by the server during the last successful anonymous check.

OpenAPI 3 service description

https://getagentdepot.com/openapi.yaml

MIME type: application/yaml

Advertised by the server during the last successful anonymous check.

Scoring methodology and Neutrality Firewall to cite

https://getagentdepot.com/api/methodology

MIME type: application/json

Advertised by the server during the last successful anonymous check.

All jobs with depotPickProviderId

https://getagentdepot.com/api/jtbds

MIME type: application/json

Advertised by the server during the last successful anonymous check.

Unscored domain packs with sourced options

https://getagentdepot.com/api/packs

MIME type: application/json

Advertised by the server during the last successful anonymous check.

Canonical contract: picks and cites, not a token router or Elo board

https://getagentdepot.com/api/canonical

MIME type: application/json

Advertised by the server during the last successful anonymous check.

Agent Benchmark Suite v2 index

https://getagentdepot.com/api/benchmark

MIME type: application/json

Advertised by the server during the last successful anonymous check.

Zero-IP telemetry policy, quarantine intake, operator contact

https://getagentdepot.com/privacy

MIME type: text/html

Advertised by the server during the last successful anonymous check.

Newest 20 releases, corrections and price changes (JSON Feed). No push: re-read each session. All channels: /api/subscribe

https://getagentdepot.com/changelog.json?limit=20

MIME type: application/feed+json