Skip to main content
Glama
dacode-dev
by dacode-dev

Agent Context API

A small paid API for coding agents that need a deterministic context preflight before sending repository text to an LLM. It counts GPT-family tokens, applies an optional model-aware budget, and redacts likely credentials.

It also exposes managed data experiments. bounty-radar normalizes live agent-work listings and preserves the difference between a canonical escrow signal and an ordinary venue listing. payanagent-health produces a fresh liveness snapshot of the ranked PayanAgent catalog. base-market-pulse combines fresh public ETH, DEX, and Base-network signals. The source code is public, but the paid value is operated aggregation, source health, bounded probing, and timestamped results that a buyer does not have to run.

agent-work-brief is the decision-oriented bundle: it combines canonical escrowed work with catalog liveness, excludes ordinary unfunded listings, and classifies opportunities that require capital before an agent spends time evaluating them.

Why pay when the source is public?

Self-hosting is deliberately allowed. The paid convenience is the maintained run: polling changing upstreams, normalizing responses, detecting partial outages, enforcing bounded work, and serving a timestamped result from a ready endpoint. A buyer can fork the code, but then owns deployment, monitoring, rate-limit handling, schema drift, and repairs. There is no claim of exclusivity; the service must earn trust through freshness and reliable operation.

Related MCP server: knitbrain

Local development

npm install
npm test
X402_FACILITATOR=https://facilitator.openx402.ai npm start

The service listens on http://localhost:8787 by default. GET /health is free. POST /v1/context-preflight returns an x402 402 Payment Required response until the caller supplies a valid Base USDC payment.

For production Base mainnet operation, use a facilitator that advertises x402Version: 2, scheme: exact, and network: eip155:8453 at its /supported endpoint; the maintained deployment uses https://facilitator.openx402.ai. The public https://x402.org/facilitator endpoint is suitable for Base Sepolia testing, not this mainnet payout configuration. Always verify current facilitator support before changing deployments.

API example

curl -X POST http://localhost:8787/v1/context-preflight \
  -H 'content-type: application/json' \
  -d '{"text":"const token = \"sk-example\";","model":"claude-sonnet","token_budget":1000}'

The default price is $0.005 per request. Set PAY_TO to change the public payout address, X402_FACILITATOR to select a compatible facilitator, and PUBLIC_HOSTS to provide a comma-separated allowlist of Worker/custom hostnames used in x402 resource URLs. Do not put private keys or facilitator credentials in this repository; the service does not need a signing key to receive payments.

Managed bounty radar

POST /v1/bounty-radar costs $0.01 through x402. Send filters such as:

{"include_unverified":false,"min_reward_usd":0,"limit":20}

The response includes source health, deadlines, claim bonds, external-spend requirements, and a payment_evidence field. Only listings explicitly marked canonical and escrowed are labeled funded; other listings remain visible only when requested and are labeled unverified.

PayanAgent catalog health

POST /v1/payanagent-health costs $0.01 through x402. It accepts an optional limit from 1 to 25 and checks the ranked public catalog using only read-only HEAD requests (falling back to OPTIONS when required). HTTP 402 means that a live payment gate responded; the service never sends payment headers or calls paid routes. The response contains per-offer status, HTTP code, latency, summary counts, and generated_at.

Base market pulse

POST /v1/base-market-pulse costs $0.01 through x402 and accepts {}. The equivalent GET route exists for agent directories and simple HTTP clients. It reads Coinbase's public ETH/USD spot endpoint, DEX Screener's public Base WETH/USDC pair feed, and public Base JSON-RPC methods for the current block and gas price. It returns per-source health, timestamped values, and partial results when one source is unavailable. It is informational data only and does not trade or submit transactions.

POST /v1/agent-work-brief costs $0.03 and accepts optional min_reward_usd, limit, and health_limit filters. It is intended for agents deciding whether current work is worth pursuing; it makes no claims, payments, or downstream paid calls.

GET /.well-known/x402 (also available as /.well-known/x402.json) publishes the machine-readable service manifest, routes, prices, network, asset, and payout address. The manifest is generated from the request host so it remains correct when the zero-cost development tunnel rotates.

When the optional agent-tools hub gateway is enabled, it forwards settled requests to /hub/v1/base-market-pulse with a hub-issued X-Hub-Secret. That upstream path returns 404 without the secret and is separate from the direct x402 routes.

MCP integration

The repository also includes a local stdio MCP server for Cline and other MCP clients:

npm run start:mcp

The preflight_context tool provides a free local tier up to 12,000 characters. Larger inputs return the paid HTTP endpoint and its x402 price. Set PAID_ENDPOINT to the currently deployed endpoint when launching the MCP server. See llms-install.md for a copy-paste client configuration.

The same server is available as a hosted, stateless Streamable HTTP MCP endpoint at https://agent-context-api-proxy.agent-context-proxy.workers.dev/mcp and is described by server.json. It is published in the official MCP Registry for client and downstream-directory discovery. This hosted route is intended for clients that can discover remote MCP servers; local execution remains available for self-hosting and offline preflight work. glama.json identifies the GitHub maintainer for directory indexing and ownership.

Design notes

  • server.js contains the Express app, x402 payment middleware, discovery metadata, and route wiring.

  • mcp-core.js defines the shared MCP tool surface; mcp-server.js runs it over stdio and /mcp runs it over Streamable HTTP.

  • analysis.js contains deterministic context analysis; bounty-radar.js contains source normalization and evidence policy.

  • payanagent-health.js contains the bounded, read-only catalog collector and probe logic.

  • market-pulse.js contains the bounded public market/RPC collectors and partial-source health output.

  • test/server.test.js covers redaction, budgets, the health route, and funded/unfunded labeling; test/payanagent-health.test.js covers health classification and the no-payment probe contract.

  • The API intentionally does not send submitted text to a third party; processing is local.

  • This is an experiment. Revenue must be verified from settlement records, not inferred from 402 responses, requests, or directory listings.

A
license - permissive license
-
quality - not tested
B
maintenance

Maintenance

Maintainers
Response time
Release cycle
Releases (12mo)
Commit activity

Resources

Unclaimed servers have limited discoverability.

Looking for Admin?

If you are the server author, to access and configure the admin panel.

Related MCP Servers

  • A
    license
    A
    quality
    A
    maintenance
    Local-first MCP server that gives any AI coding agent per-project memory, workflow intelligence, and always-on, lossless token & context optimization.
    37
    31
    3
    MIT
  • A
    license
    -
    quality
    C
    maintenance
    MCP server that reduces AI agent token usage by up to 90% through intelligent context compression. Enables efficient code exploration, multi-file refactoring, and debugging by providing tools for smart reading, searching, and managing code context.
    4
    MIT

View all related MCP servers

Related MCP Connectors

  • Hosted MCP server for LLM cost estimation, model comparison, and budget-aware routing.

  • A paid remote MCP for OpenAI Codex context compressor, built to return verdicts, receipts, usage log

  • A paid remote MCP for OpenAI Codex agent coordination MCP, built to return verdicts, receipts, usage

View all MCP Connectors

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/dacode-dev/agent-context-api'

If you have feedback or need assistance with the MCP directory API, please join our Discord server