Plith

Five tools for AI agents: deduplication, cost estimation, output validation, behavioral governance, and cross-org failure learning. Pick your integration path below.

Keys are free — no signup form, no dashboard. Just one API call. Both fields are optional but recommended.

curl -X POST https://plith.ai/api/keys \ -H "Content-Type: application/json" \ -d '{"email": "you@company.com", "name": "Your Org"}'

{ "org_id": "org_abc123", "key": "plth_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx", "prefix": "plth_xxxxxx", "message": "Store this key securely — it will not be shown again." }

⚠The key is returnedonceand cannot be retrieved again — save it immediately.

•Free tier:20,000 credits on signup, granted once you verify your email — no card required.

•Rate limit:3 keys per IP per 24 hours.

Add Plith to your MCP client config. Plith exposes 14 MCP tools usingproduct_actionnaming — e.g.,dedupq_check,burnrate_estimate. Arguments are passed directly, noaction/paramsnesting needed.

curl -X POST https://plith.ai/api/keys \ -H "Content-Type: application/json" \ -d '{"email": "you@example.com"}'

# Response — store the key securely, it is shown once { "org_id": "org_abc123", "key": "plth_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx", "prefix": "plth_xxxxxx", "message": "Store this key securely — it will not be shown again." }

⚠The key is shownonceand never stored in plaintext. Copy it now.

For Claude Desktop, edit~/Library/Application Support/Claude/claude_desktop_config.json. For Claude Code, runclaude mcp addor edit.mcp.jsonin your project root.

{ "mcpServers": { "plith": { "url": "https://plith.ai/api/mcp/rigor", "headers": { "x-api-key": "YOUR_API_KEY" } } } }

✓Restart your MCP client. The agent now has access to 14 Plith MCP tools:dedupq_check,burnrate_estimate,qualitygate_validate,guardrail_check,pitfalldb_query, and more.

The tool name uses underscore naming (e.g.,dedupq_check); arguments are passed directly as the tool's input parameters.

# dedupq_check — check for duplicate tasks { "name": "dedupq_check", "arguments": { "content": "Summarize the Q3 earnings report", "task_id": "task_789" } } # burnrate_estimate — estimate LLM cost { "name": "burnrate_estimate", "arguments": { "plan": [ {"step": 1, "provider": "anthropic", "model": "claude-sonnet-4-6", "estimated_input_tokens": 2000, "estimated_output_tokens": 500} ] } } # Legacy: flat names (dedupq, burnrate, etc.) and "plith" umbrella # still work for backward compatibility { "name": "plith", "arguments": { "action": "dedupq.check", "params": { ... } } }

Pass your API key in thex-api-keyheader on every request. Only/api/keysis unauthenticated.

curl https://plith.ai/api/dedupq/check \ -H "x-api-key: plth_xxxxxxxxxxxx" \ -H "Content-Type: application/json" \ -d '{...}'

All product endpoints requirex-api-keyheader except/api/keys.

/api/keysCreate org + API key. Body: { email?, name? }

/api/keys/verifyVerify an API key is valid

/api/guardrail/policiesCreate a guardrail policy. Body: { name, rules }

/api/dedupq/checkCheck for duplicate or semantically similar cached tasks.

/api/dedupq/completeMark a task complete and store its result in the dedup cache.

/api/burnrate/estimateEstimate LLM cost for a multi-step agent plan.

/api/burnrate/trackRecord actual LLM spend for a completed task.

/api/burnrate/optimizeGet cheaper model substitutions for a plan given a target budget.

/api/burnrate/budgetGet current period spend, budget, and burn rate.

/api/qualitygate/validateValidate agent output against configured quality rules.

/api/qualitygate/trendsGet validation pass/fail trends over time.

/api/guardrail/checkCheck a proposed action against behavioral policies.

/api/guardrail/policiesList your active guardrail policies.

/api/pitfalldb/queryQuery known failure patterns for a task type.

/api/pitfalldb/reportReport a new agent failure. Contributes to the cross-org database.

/api/pitfalldb/statsDatabase stats: total pitfalls, top task types, recent additions.

/api/billing/balanceGet credit balance and plan details.

/api/billing/checkoutPurchase credits. Amounts: 10, 25, or 100 USD.

No card required.Create a keyand start building.

Agent Mail Rooms is a paid remote MCP endpoint for OpenAI Codex agent coordination MCP. It exposes structured JSON tools, a public server card, token-based access, usage receipts, and audit-