API Platform
Deterministic tools your agent can actually call
17 small HTTP endpoints for AI builders and delivery teams. Almost none of them run a language model, so the same input returns the same output every time and a call costs about as much as the arithmetic it performs.
Try any endpoint straight from the reference — it has a request console that calls the API from your browser.
Or skip the HTTP entirely
Every endpoint is also an MCP tool. Point Claude, Claude Code, or any MCP host at one URL and the whole toolset appears natively — no client code, no tool definitions to maintain, nothing to redeploy when we ship a new endpoint.
claude mcp add --transport http quasslabs https://api.quasslabs.com/mcp \
--header "Authorization: Bearer qlk_YOUR_API_KEY"Auth, rate limiting and credit metering behave exactly as they do over REST, and the gateway is not itself metered — going through MCP never costs more than calling directly.
The 17 endpoints
Prices are per successful call. A rate-limited request is never billed, and anything that fails with a server error is refunded automatically.
Writing and prompts
Detect and remove AI-writing tells, and grade a system prompt before it ships.
-
/v1/humanizer-score1 credit Detect AI-writing tells in text MCP:humanizer_score -
/v1/humanize1 credit Rewrite text to remove AI-writing tells MCP:humanize -
/v1/alignment-gate2 credits Run an AI system prompt (and optional agent config) through a 45-criterion alignment/safety rubric MCP:alignment_gate
Engineering decisions
Pick a model, price it, and check what your dependencies and logs are hiding.
-
/v1/model-advisor1 credit LLM selection + cost advisor MCP:model_advisor -
/v1/dependency-check1 credit Dependency-manifest hygiene check MCP:dependency_check -
/v1/log-analyzer2 credits Analyze a log file MCP:log_analyzer -
/v1/log-audit-creator2 credits Generate a concrete audit-logging spec for an app MCP:log_audit_creator
Delivery and QA
Lint the work before it reaches a sprint - tickets, criteria, coverage, bugs.
-
/v1/ticket-lint1 credit Full Jira-ticket quality linter (whole ticket, not just AC - use acceptance_criteria_validator for the deep AC pass) MCP:ticket_lint -
/v1/acceptance-criteria-validator1 credit QA advisor: validate acceptance criteria quality MCP:acceptance_criteria_validator -
/v1/coverage-check1 credit Scope-completeness advisor MCP:coverage_check -
/v1/defect-triage1 credit Bug-triage advisor MCP:defect_triage -
/v1/sprint-health1 credit Sprint/delivery health advisor MCP:sprint_health -
/v1/worklog-guardrail1 credit Worklog/timesheet policy checker - run a batch of worklog entries through configurable guardrails BEFORE posting them to a time-tracking system MCP:worklog_guardrail
Everything else
Diagram accessibility, SaaS health maths, public-land lookup, and file uploads.
-
/v1/diagram-contrast1 credit Diagram quality QA MCP:diagram_contrast -
/v1/saas-metrics1 credit SaaS health advisor MCP:saas_metrics -
/v1/blm-land-lookup1 credit Look up Bureau of Land Management land status for a US coordinate: BLM ownership, jurisdiction, allowed activities (rockhounding, camping), permit requirements, and special designations (ACEC, Wilderness, WSA, NM/NCA) MCP:blm_land_lookup -
/v1/uploadsfree Get a presigned S3 PUT URL to upload a file (policy doc, config, log, up to 5MB) that a file-ingesting endpoint then reads by upload_id MCP:uploads
How billing works
Predictable
- Rate limits are checked before credits, so a
429costs nothing. - The decrement is atomic and conditional — if your balance is short you get a
402and the endpoint does not run. - Server errors are refunded automatically.
Known gap
- There is currently no way to read your credit balance — no endpoint returns it and no header carries it.
- Today you find out via
402 insufficient_credits. Treat it as an alert, not something to retry.
Full detail in errors and limits, and the machine-readable spec is at openapi.json.