askround
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| ASKROUND_DIR | No | Directory where rounds are saved. Defaults to .askround. | |
| GEMINI_API_KEY | No | API key for the Gemini provider. Can be used instead of GOOGLE_API_KEY. | |
| GOOGLE_API_KEY | No | API key for the Gemini provider. Can be used instead of GEMINI_API_KEY. | |
| OPENAI_API_KEY | No | API key for the OpenAI provider. | |
| ASKROUND_CONFIG | No | Path to the askround config file. Defaults to %APPDATA%\askround\config.toml on Windows or ~/.config/askround/config.toml elsewhere. | |
| ASKROUND_PRICES | No | Path to a JSON file that replaces the default price table. | |
| ASKROUND_PRIVATE | No | List of private names to detect in questions. | |
| ANTHROPIC_API_KEY | No | API key for the Anthropic provider. | |
| ASKROUND_MAX_COST | No | Maximum estimated cost allowed for a round. | |
| PERPLEXITY_API_KEY | No | API key for the Perplexity provider. | |
| ASKROUND_GEMINI_MODEL | No | Override the model used for the Gemini provider. | |
| ASKROUND_OPENAI_MODEL | No | Override the model used for the OpenAI provider. | |
| ASKROUND_PRIVATE_FILE | No | Path to a file containing private names to detect in questions. | |
| ASKROUND_EXTRACT_MODEL | No | Override the model used for extraction passes. | |
| ASKROUND_ANTHROPIC_MODEL | No | Override the model used for the Anthropic provider. | |
| ASKROUND_PERPLEXITY_MODEL | No | Override the model used for the Perplexity provider. |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
Ask other AI assistants a question and read what they recommend. Start with askround_doctor to see which providers can run; askround_ask returns a report, and askround_answer one answer's text.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| askround_askA | Ask one question to several AI assistants (OpenAI, Gemini, Anthropic, Perplexity) with web search, each several times, and read what they recommend. Returns a brief of a few hundred tokens: for each product the runs naming it and its best position per provider, its top claims and the rivals, each fact with the ids of the answers it came from ('openai#2'). Read an answer, or the passage behind a fact, with askround_answer; detail: true returns the full report instead. Refused before any call if the question names a product it tests, carries personal data, or is estimated above max_cost. |
| askround_answerB | One answer from a round already asked, by the id a brief gives ('openai#2'): its full text and sources, or with |
| askround_runC | Ask every question of a product file (a prompt set) and return a brief per question: the runs naming the product and its best position per provider, its claims, the rivals, each fact with its answers' ids. detail: true returns every question's full report instead. |
| askround_jobB | A round or prompt set started with background: true - running (with how many answers are in), done (with its brief or report), refused, failed, or interrupted (with the answers it finished, which are kept). |
| askround_researchA | Ask a research question to every assistant with web search and merge the answers: the claims they make, how many providers support each, each provider's passage, and the sources they cited with primary ones (repositories, licence texts, papers, official docs) first. Support is agreement among assistants, not proof. |
| askround_reportB | The report of a stored round or prompt set (default: the latest). |
| askround_diffB | What changed between two runs of the same questions (default: the latest two): share and rank, rivals, cited pages, claims. |
| askround_doctorA | Which providers can run here: the key variable each reads (never its value), the model and the search tool. Makes a free check call to each. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 8 tools
The three query tools (ask, research, run) all send questions to multiple assistants, but they are differentiated by input shape (single question vs. research question vs. prompt file) and output shape (ranked brief vs. merged claim support vs. per-question briefs). Retrieval tools (answer, report, job) are clearly layered by identifier vs. full report vs. async status, though ask vs. research remains the one boundary an agent could misjudge. Overall the set is mostly distinct with descriptions carrying the distinction.
All eight tools follow a single predictable pattern: lowercase askround_ prefix plus one short word (ask, answer, report, run, job, research, diff, doctor). The only minor deviation is mixing verbs (ask, run, research, diff) with nouns (answer, report, job, doctor) rather than a strict verb_noun scheme, but the style itself is uniform.
Eight tools is well within the well-scoped range, and each one maps to a distinct step of the workflow: query, batch query, research merge, answer lookup, report retrieval, async polling, change detection, and provider diagnostics. No tool appears redundant or filler.
The surface covers the full lifecycle of asking, retrieving (answer/report), batching (run), async execution (job), comparison (diff), and environment checking (doctor). Minor gaps exist: there is no tool to list stored rounds or prompt sets (report defaults to latest, requiring an id from prior output), and no job cancellation or cost-estimate tool despite max_cost refusal behavior.