opencode-specstream 0.1.1 → 0.1.4

package/README.md

@@ -1,11 +1,16 @@
 # opencode-specstream
 
-SpecStream plugin for [OpenCode](https://opencode.ai). SpecStream is the live web data engine for teams and agents: describe a list in plain words ("roofing companies in Columbus, Ohio with their website and phone") and get every matching row, each with the page it came from. You see the price first, you approve a ceiling, and You pay for each result delivered. A search that finds nothing costs nothing.
+SpecStream plugin for [OpenCode](https://opencode.ai). SpecStream is the live web data engine. One platform to build, search, extract, enrich, map, and monitor data across the entire web.
+
+Ask for any dataset in plain words, and SpecStream builds it from the entire live web: every match found, extracted and enriched into structured records, then monitored for change.
+
+- Source URL and read time on every value.
+- You see the price first and approve a ceiling. You pay for each row delivered, and a run that finds nothing costs nothing.
 
 The plugin does two things when OpenCode starts:
 
 - adds the hosted SpecStream MCP server, `https://api.specstream.ai/mcp`, to your config as `specstream`
-- adds four agent skills: `compile` (read the request back), `review-work` (get the price), `approve-work` (approve it) and `continue-work` (read the rows, send them on, keep them current)
+- adds six agent skills: `search-web` (Search API), `extract-pages` (Extract API), `compile` (show how SpecStream read the request), `review-work` (get the price), `approve-work` (approve it) and `continue-work` (read the rows, send them on, keep them up to date)
 
 ## Install
 
@@ -37,7 +42,7 @@ The server uses OAuth, so you do not paste a key. Run:
 opencode mcp auth specstream
 ```
 
-Your browser opens, you sign in to SpecStream, and OpenCode keeps the token. New accounts start with $10 of work free for 30 days, no card.
+Your browser opens, you sign in to SpecStream, and OpenCode keeps the token. New accounts start with $10 of credit for 30 days, no card.
 
 To use an API key instead, for example on a machine with no browser, create one under **Agents & API** in the [workspace](https://specstream.ai/workspace) and set it where OpenCode starts:
 
@@ -49,11 +54,7 @@ With the variable set, the plugin sends it as a bearer header and turns the brow
 
 ## Use it
 
-Ask for a list:
-
-> Find food banks in Texas, with their annual revenue.
-
-The agent reads the request back, shows the quoted ceiling, and starts only after you say yes. Rows come back with their source pages, ready to export as CSV, Excel or JSON or to send to Google Sheets, Notion, Airtable, HubSpot or Salesforce.
+Ask in plain words. The agent shows how SpecStream read the request and the quoted ceiling, and starts only after you say yes. Rows come back with their source pages, ready to export as CSV, Excel or JSON or to send to Google Sheets, Notion, Airtable, HubSpot or Salesforce.
 
 ## Options
 

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "opencode-specstream",
-  "version": "0.1.1",
-  "description": "SpecStream plugin for OpenCode: adds the SpecStream MCP server and four agent skills, so your agent can ask for a list in plain words and get every matching row with its source.",
+  "version": "0.1.4",
+  "description": "The live web data engine. One platform to build, search, extract, enrich, map, and monitor data across the entire web.",
   "type": "module",
   "main": "index.ts",
   "files": [

package/skills/compile/SKILL.md

@@ -9,6 +9,6 @@ SpecStream is the live web data engine for teams and agents. Use the configured
 
 Use `compile` to read the person's request. Say back what it found: the kind of thing, the conditions, the details to bring back and the quantity. Separate the conditions that must hold from the details that are nice to have. Use `edit_spec` for a deliberate change. Never drop a hard condition quietly, and never treat missing evidence as a no. Reading a request does not start any work.
 
-Suggest requests that work well: local businesses with ratings, companies from registries, grants, tenders, clinical trials, papers, datasets, regulations, vulnerabilities, apps, repositories, packages, podcasts and books. Companies that use a tool, funding and new-executive signals, US nonprofits and UK financial firms by FCA status also work. If someone asks for people by job title, say that results may be few. `list_request_templates` returns requests that already work for each kind.
+If someone asks for people by job title, say that results may be few. `list_request_templates` returns requests that already work for each kind.
 
 The same saved request can be revised, branched, compared and shared, in the workspace or through this agent. Never ask for an API key in chat. Use the connector's own sign-in.

package/skills/extract-pages/SKILL.md

@@ -0,0 +1,10 @@
+---
+name: extract-pages
+description: Read the main text of up to 20 web pages, PDFs, Word or Excel files live with the Extract API ($0.80 per 1,000 pages).
+---
+
+# Read pages and documents live
+
+SpecStream is the live web data engine for teams and agents. Use the configured MCP server at `https://api.specstream.ai/mcp`.
+
+Use `read_pages` with `{"urls": ["<url>", ...]}` (1 to 20 `http` or `https` URLs) to read the clean main text, title and published date of web pages, PDFs, Word documents or Excel spreadsheets. Each page that returns text costs $0.0008 ($0.80 per 1,000 pages) from the account balance, and a page with no text is not charged. No quote or approval is needed.

package/skills/search-web/SKILL.md

@@ -0,0 +1,12 @@
+---
+name: search-web
+description: Answer a question from the live web in seconds with the Search API ($5 per 1,000 searches, no quote needed).
+---
+
+# Search the live web
+
+SpecStream is the live web data engine for teams and agents. Use the configured MCP server at `https://api.specstream.ai/mcp`.
+
+Use `search_web` with `{"q": "<the question>"}` when the person needs one fact, a fast web search with sources, or a quick look at the first rows of a request before pricing a full Deep Search run. Set `"passages": true` when you want the main text passages of the closest pages included in the reply.
+
+Each call costs $0.005 ($5 per 1,000 searches) from the account balance and needs no quote or approval. The reply returns `kind` (`fact` or `list`), `answer` (with `value`, `page` and `seen_at`), `rows` and ranked `sources`. When `kind` is `list`, `handoff` gives the estimated size and the `compile` call to price and run the full dataset.