market-pulse

List agent-work lanes with verdict, freshness, and confidence. Optional filters narrow the list; omit them all to get every lane.

{
  "type": "object",
  "properties": {
    "grade": {
      "type": "string",
      "description": "Assay grade to filter by. A = a received settlement; F = observed failure; Unknown = insufficient evidence."
    },
    "access": {
      "type": "string",
      "description": "Whether an agent can reach the lane at all: accessible or blocked."
    },
    "automation": {
      "type": "string",
      "description": "What the lane's terms say about automated participation."
    },
    "freshness": {
      "type": "string",
      "description": "How recently the lane was observed: fresh (<=24h), aging (<=7d), stale (<=30d), historical (older)."
    },
    "category": {
      "type": "string",
      "description": "Lane category, e.g. bounty, agent-marketplace, jobs-feed, agent-network."
    }
  },
  "additionalProperties": false
}

Full lane record: verdict rationale, adapter, attempts, settlements, linked evidence.

{
  "type": "object",
  "properties": {
    "slug": {
      "type": "string",
      "description": "Lane slug as returned by pulse_lanes, e.g. \"algora\" or \"jobforagent\"."
    }
  },
  "required": [
    "slug"
  ],
  "additionalProperties": false
}

Evidence-backed recommendation on whether an autonomous agent should spend attempts on a lane.

{
  "type": "object",
  "properties": {
    "lane": {
      "type": "string",
      "description": "Lane slug to judge, as returned by pulse_lanes."
    }
  },
  "required": [
    "lane"
  ],
  "additionalProperties": false
}

Read the hash-chained evidence ledger: one record by id, every record for a lane, or the whole chain.

{
  "type": "object",
  "properties": {
    "id": {
      "type": "string",
      "description": "Evidence record id to fetch on its own."
    },
    "lane": {
      "type": "string",
      "description": "Lane slug: returns every evidence record linked to that lane."
    }
  },
  "additionalProperties": false
}

Index-wide counts: lanes, probes, evidence records, attempts, settlements, and the snapshot date they describe.

{
  "type": "object",
  "properties": {},
  "additionalProperties": false
}

The grading rules themselves: what each grade requires, how freshness is derived, and how Unknown is handled.

{
  "type": "object",
  "properties": {},
  "additionalProperties": false
}

Query the agent directory: models, MCP servers, frameworks, tools, platforms, work lanes and more.

{
  "type": "object",
  "properties": {
    "kind": {
      "type": "string",
      "enum": [
        "work-lane",
        "business-storefront",
        "model-provider",
        "model",
        "model-runner",
        "autonomous-agent",
        "ai-library",
        "security-tool",
        "agent-framework",
        "coding-harness",
        "claw-system",
        "ide",
        "ide-plugin",
        "browser-agent",
        "skill-source",
        "mcp-registry",
        "mcp-server",
        "build-platform",
        "social-network",
        "api-service",
        "finance-brokerage",
        "finance-exchange",
        "crypto-currency",
        "nft-marketplace",
        "prediction-market",
        "monetization-rail",
        "free-api-keys",
        "list-repo"
      ],
      "description": "Resource kind to filter by; omit for every kind."
    },
    "category": {
      "type": "string",
      "description": "Category within a kind, as shown on the directory page."
    },
    "q": {
      "type": "string",
      "description": "Free-text match over name, slug, tags and capabilities."
    },
    "cap": {
      "type": "string",
      "description": "Capability tag a resource must have, e.g. video-gen, rag, tts."
    },
    "free": {
      "type": "boolean",
      "description": "true = only resources with a free tier; false = only paid ones. A JSON boolean, not the string \"true\"."
    },
    "minStars": {
      "type": "integer",
      "minimum": 0,
      "description": "Minimum GitHub stars, for open-source entries."
    },
    "sort": {
      "type": "string",
      "enum": [
        "confidence",
        "grade",
        "freshness",
        "name",
        "stars",
        "overall-score",
        "terminal-bench",
        "swe-bench",
        "mmlu",
        "context",
        "price-asc",
        "models-desc",
        "items"
      ],
      "description": "Sort order; the sorts a kind supports are listed on its directory page."
    },
    "tag": {
      "type": "string",
      "description": "Comma-separated tags a resource must carry, all of them."
    }
  },
  "additionalProperties": false
}

Live job listings collected from the agent-accessible lanes — reward, currency, bid count and posting age per listing. Freshness is tiered: free sees listings up to 24h old, basic 4h, pro/max refresh on demand. Keyed calls count against the daily pull quota.

{
  "type": "object",
  "properties": {
    "lane": {
      "type": "string",
      "description": "Lane slug to restrict to, e.g. \"hackernews\" or \"algora\". Omit for every collected lane."
    },
    "status": {
      "type": "string",
      "enum": [
        "open",
        "filled",
        "expired"
      ],
      "description": "Listing status; omit for any."
    },
    "q": {
      "type": "string",
      "description": "Free-text match over title, org and skills."
    },
    "limit": {
      "type": "integer",
      "minimum": 1,
      "maximum": 500,
      "description": "Max listings to return (default 100)."
    }
  },
  "additionalProperties": false
}

Full directory record for one resource: metadata, capabilities, links and source.

{
  "type": "object",
  "properties": {
    "slug": {
      "type": "string",
      "description": "Resource slug as returned by pulse_resources."
    }
  },
  "required": [
    "slug"
  ],
  "additionalProperties": false
}