Agentic Index
Capy vs Tembo (2026)
Both vendors last verified
Capy is a coding agent you hand work to, and Tembo is a platform that runs the coding agents you already use. That verdict is the Agentic Index coverage score, graded from each vendor's own published materials.
Both start each task in an isolated cloud machine with a browser and a desktop, take work from Slack, Linear and GitHub, and finish in a pull request that neither merges unless a person says so.
Capy, from Scrapybara, Inc., is its own AI software engineer. Each thread runs on a Firecracker microVM, splits work across subagents, and lets a person pick the model per thread from OpenAI, Anthropic, Google, xAI and others, or run one task on several models to compare.
Tembo ships no agent of its own. It runs Claude Code, Codex, Cursor, OpenCode, Amp and Pi across repositories, starts them from events in GitHub, Linear, Jira, Sentry and Slack or on a schedule, and can self host on AWS, Azure or GCP in early access.
For a solo developer the entry prices differ. Capy Lite is $20 a month with $20 of credits. Tembo Pro is $60 a month as a usage allowance for up to five users.
Choose Capy to delegate tasks to one agent with a choice of models. Choose Tembo to run the agents your team already pays for on triggers and schedules.
On the Agentic Index coding agent ranking, neither Capy nor Tembo clears the bar, which asks for all five merge loop capabilities documented in full. Capy does not document knowledge grounding and RAG in full, nor testing, debugging and optimization; Tembo documents two of the five in full. 3 of the 96 vendors in the lane clear it. See the coding agent ranking
This comparison is published by Agentic Index, an independent agentic AI vendor research platform. Capy and Tembo are each graded against the same 14 capability Agentic Index taxonomy, from the vendor's own public materials under the Agentic Index verification standard, alongside 1,165 researched vendors. No vendor pays for placement and no vendor has reviewed this page. How this evidence is graded
Choose Capy if
- You work alone and want a low entry price. Capy Lite costs $20 a month ($16 billed annually) with $20 of credits, and there is no per seat fee.
- You want to pick the model per task. Each thread can run on models from OpenAI, Anthropic, Google, xAI, Moonshot, DeepSeek and others, and one task can run on several models to compare approaches.
- Large tasks need splitting up. Capy delegates to subagents that nest up to three levels, each on its own model, and keeps a plan you can follow step by step.
- You sometimes want the agent on your own machine. Capy's desktop app runs agents on your Mac or PC as well as in its cloud VMs in US East.
- You already pay for another coding plan. Capy can connect existing Codex, Copilot, SuperGrok or Claude subscriptions, which skip the token charge.
Choose Tembo if
- You already use Claude Code, Codex, Cursor or OpenCode. Tembo runs those agents, plus Amp and Pi, in cloud VMs preloaded with your dependencies, and the agent and model stay your choice.
- Work should start from events rather than prompts. Runs fire from GitHub, GitLab, Bitbucket, Linear, Jira, Sentry, Slack and HubSpot, on cron schedules, from webhooks or from an issue assigned to Tembo.
- One change spans several repositories. Tembo coordinates the change across repositories in one pass, with one pull request per repository.
- Recurring chores eat your week. Twelve agent templates cover jobs such as changelogs, standup summaries, test pruning and Sentry triage into Slack.
- Your security team wants the platform in your own cloud. Tembo self hosts as a NixOS image on AWS, Azure or GCP in early access, and lists SOC 2, ISO 27001, ISO 42001, GDPR and HIPAA with reports under NDA.
| Feature | C Capy |
T Tembo |
|---|---|---|
| Action & orchestration | ||
|
Integrations & Tool Calling Ability to connect agents to real systems through native integrations, OAuth-authenticated actions, custom tools, APIs, webhooks, or MCP-compatible tools. |
||
|
CapyIntegrations & Tool Calling Capy acts inside the tools a team uses. A GitHub App opens and reviews pull requests, Slack threads take @Capy, delegating a Linear issue creates a native agent session that streams progress back, and the home page says it imports Sentry reports. MCP servers connect over HTTP or stdio with OAuth, Tailscale ephemeral nodes reach private databases and internal services, Vercel previews and Paper are integrated, and OIDC federation gives machines short lived GCP, AWS, Azure or Vault credentials. GitHub Enterprise Server is not supported. Sourcedocs.capy.ai/integrations/mcp, integrations/github, integrations/linear, integrations/tailscale and capy.airead 2026-10-28 |
||
|
TemboIntegrations & Tool Calling Named integrations cover GitHub, GitLab and Bitbucket for source control, Linear, Jira and Confluence for tickets and docs, Slack and Microsoft Teams for chat, Sentry for errors, and Notion, HubSpot, Snyk, BigQuery, AWS and Tailscale, plus a catalog of about forty MCP servers and custom MCP servers over HTTP, SSE or stdio. The same connections assemble agent context and return pull requests as output. Datadog is reachable as an MCP query integration only, not as a trigger, although the homepage lists it among trigger sources. Environments are preloaded with the customer's own dependencies and tools so agents reach the real toolchain. Sourcedocs.tembo.io/integrations, docs.tembo.io/integrations/datadog and tembo.ioread 2026-10-28 |
||
|
Workflow Orchestration Ability to sequence, branch, retry, route, and combine deterministic workflow nodes with autonomous agent steps. |
||
|
CapyWorkflow Orchestration A thread delegates work to subagents that nest up to three levels, on a shared or a fresh machine, each with its own model, and the home page describes hundreds of parallel agents, each task on its own branch and VM. Captains, an opt in experiment, direct crews of threads and keep their own Drive, and tasks were renamed subagents in October. Sourcedocs.capy.ai/subagents, docs.capy.ai/captains, capy.ai and capy.ai/changelog 0.4.3read 2026-10-28 |
||
|
TemboWorkflow Orchestration Tembo is a managed control plane that runs other vendors' coding agents, Claude Code, Codex, Cursor, OpenCode, Amp, Pi and fx, as background agents, dozens at once, each session on its own VM. A multi repository session opens one pull request per repository, runs can be scheduled on a cadence, related triggers continue one session, and sessions can be forked. Tembo positions it as the layer above individual agents rather than one agent in one session, and no sub agent mechanism of its own is documented. Sourcedocs.tembo.io/features/agents, tembo.io/agents and tembo.io/changelog 2 October 2026read 2026-10-28 |
||
|
Triggers & Channel Coverage How agents wake up and where they work: schedules, webhooks, message events, CRM events, inbox events, chat, email, voice, and collaboration tools. |
||
|
CapyTriggers & Channel Coverage Automations carry one to twenty triggers: cron with a five minute floor, GitHub events, Slack message bursts, an incoming webhook and on demand runs from the API, and the API adds Linear triggers (issue created, status changed, end of cycle) and Sentry triggers. Delegating a Linear issue or mentioning @capy on a pull request comment starts work, and a thread wakes on its own when CI finishes, a review lands or a PR merges. Sourcedocs.capy.ai/automations, docs.capy.ai/openapi.json, integrations/linear and capy.ai/changelog 0.4.2read 2026-10-28 |
||
|
TemboTriggers & Channel Coverage Agent triggers fire on events from GitHub, GitLab, Bitbucket, Jira, Linear, Sentry (issue created), Slack (message or reaction), Confluence and HubSpot, on cron schedules in UTC, and on a webhook or API POST to the runs endpoint. A developer can also mention @tembo in a thread, use a macro, or assign a Linear, Jira or GitHub issue to Tembo and receive a pull request back. Datadog is not a trigger source. Sourcedocs.tembo.io/api/v1/agents/create-an-agent-trigger, docs.tembo.io/integrations/linear and tembo.io/automationsread 2026-10-28 |
||
| Knowledge & context | ||
|
Knowledge Grounding & RAG Ability to ground agent behavior in company data through document ingestion, retrieval, external knowledge APIs, semantic search, or RAG layers. |
||
|
CapyKnowledge Grounding & RAG Agents work over a real checkout of the customer's repository inside their VM, nested AGENTS.md files load when a folder is first touched, a Review context page holds organization rules and files, and imported context such as Sentry reports and attached files lands on the agent's machine. No persistent codebase index, embeddings or retrieval layer is described. Sourcedocs.capy.ai/instructions, docs.capy.ai/review and capy.ai/changelog 0.3.0read 2026-10-28 |
||
|
TemboKnowledge Grounding & RAG Projects snapshot cloned repositories with dependencies installed so agents start with the codebase ready, and rule files (tembo.md, AGENTS.md, CLAUDE.md and others) are read at the start of each session. Context for a run is assembled from the repositories plus Linear, Jira, Sentry, Slack and GitHub. Tembo documents no codebase index, knowledge graph, embeddings layer or retrieval surface of its own; deep codebase understanding within a run comes from the hosted agent. Sourcedocs.tembo.io/features/projects, docs.tembo.io/features/rule-files and docs.tembo.io/integrationsread 2026-10-28 |
||
|
Memory & State Persistence Ability to persist context across a run, conversation, workflow, user, team, or longer-term memory layer. |
||
|
CapyMemory & State Persistence A thread is a durable conversation stored on Capy's servers that holds the full history of its messages, machines, subagents and PRs, survives archiving and resumes when woken. An opt in Memory experiment keeps what a user teaches Capy across threads and review findings list the memory they rest on, but no docs page states its scope, lifetime or a way to delete entries. Drive holds AGENTS.md instructions at personal or organization level, which are instructions rather than memory. Sourcecapy.ai/changelog 0.3.0 and 0.5.0, docs.capy.ai/drive and docs.capy.ai/async-workread 2026-10-28 |
||
|
TemboMemory & State Persistence Project snapshots persist the customer's dependencies and tooling across runs so agents do not rebuild the environment each time, related triggers continue one session, and sessions can be forked. The enterprise page says session history powers memory that compounds over time, but the docs describe no memory page, scope, lifetime or purge path, and scheduled automations keep state only through getAgentState and setAgentState tools. Any cross session memory sits with the hosted agent or a third party MCP server such as Supermemory. Sourcedocs.tembo.io/features/projects, tembo.io/enterprise and tembo.io/changelog 21 November 2025read 2026-10-28 |
||
| Control & trust | ||
|
Human Oversight & Guardrails Approval steps, consent checkpoints, escalation rules, structured guardrails, policy constraints, and pause/resume controls. |
||
|
CapyHuman Oversight & Guardrails Capy never merges or enables auto merge unless asked in the thread, the Merge menu offers Enable auto merge and Force merge, and a root AGENTS.md can grant standing merge permission. The agent waits for a go ahead when told to, review mode runs Off, Once or on every push, paired devices ask before use, any agent can be stopped mid run, and a thread or Captain tags the user under Needs you when blocked on a decision, access or credentials. Sourcedocs.capy.ai/admin/security, docs.capy.ai/pull-requests, docs.capy.ai/async-work and capy.ai/changelog 0.6.0read 2026-10-28 |
||
|
TemboHuman Oversight & Guardrails Every run ends with a draft pull request or an artifact the team reviews, and the review page says nothing merges until a person makes the call. A planning mode was added in September 2026 and the API accepts a normal or planning mode, and since October a browser notification fires when an agent needs approval, but no docs page describes an approval gate, a permission policy or per tool control, and the GitHub integration docs say agents can merge or close pull requests through tool calls. The per team, repository or agent approval gates appear only on the enterprise page. The gate is the customer's own pull request review. Sourcetembo.io/review, tembo.io/changelog 9 September and 9 October 2026, docs.tembo.io/integrations/github and tembo.io/enterpriseread 2026-10-28 |
||
|
Security, Identity & Governance RBAC, SSO, auditability, encryption, least-privilege tool access, compliance posture, and data handling policy. |
||
|
CapySecurity, Identity & Governance Capy announced SOC 2 Type II on 10 March 2026, with the report available through its trust center at trust.capy.ai. Organizations carry Owner, Admin and Member roles, Enterprise adds SAML and OIDC single sign on for Okta, Azure AD and Google Workspace, audit logging and zero data retention by agreement, the desktop app signs in with SSO, and the enterprise page claims TLS 1.3 and AES-256. No SCIM is documented. Sourcecapy.ai/blog/capy-achieves-soc-2-type-ii, capy.ai/enterprise and docs.capy.ai/admin/membersread 2026-10-28 |
||
|
TemboSecurity, Identity & Governance Tembo's security and enterprise pages list SSO over SAML 2.0 and OIDC with Okta, Azure AD, Google Workspace or any compliant IdP, role based access at team, repository and agent level, API keys with scoped permissions, expiry, rotation and revocation, AES-256 at rest and TLS 1.3 in transit, and SOC 2 Type II, ISO 27001, ISO 42001, GDPR and HIPAA with reports under NDA. The vendor's own surfaces disagree on the attestation: the security page shows both a Type II claim and a Type I badge, and the trust center lists SOC 2 Type 1 and Type 2 behind a request while its self assessment says the company is working on its security compliance, so the attestation type is not relied on. The docs carry no SSO or SAML page, only organization member roles through the API. Sourcetembo.io/security, tembo.io/enterprise, trust.tembo.io and docs.tembo.io/apiread 2026-10-28 |
||
|
Observability & Auditability Traces, logs, execution histories, metrics, audit events, and debugging detail for production agent behavior. |
||
|
CapyObservability & Auditability Each thread keeps its full transcript and tool output, shows the agent's plan, terminal output, file changes and diffs, and a status (active, waiting, needs attention, ready for review), with the machine's desktop streamed live into the thread pane. An Activity page and usage reports break spend down by member, model and thread, the API streams events over SSE, end to end test runs record annotated video, and audit logging is on the Enterprise plan. Sourcedocs.capy.ai/async-work, docs.capy.ai/machines, docs.capy.ai/api-reference/overview and capy.ai/enterpriseread 2026-10-28 |
||
|
TemboObservability & Auditability Every agent session has a centralized audit trail with configurable retention (90 days by default) and SIEM export on the security page, and the enterprise page says each prompt, agent action, file change and approval is logged. In the docs, session events and agent runs are retrievable through the API, Insights reports usage by provider, and a transcript export added on 23 October 2026 gives the full tool call record of a session as Markdown or JSON with secrets scrubbed. Sourcetembo.io/security, tembo.io/enterprise, docs.tembo.io/api and tembo.io/changelog 23 October 2026read 2026-10-28 |
||
|
Deployment & Data Residency Deployment modes and options, including SaaS, dedicated cloud, VPC, on-prem, hybrid, local runtime, and self-hosting. |
||
|
CapyDeployment & Data Residency Each cloud thread runs on its own Firecracker microVM on Ubuntu 24.04, sleeping after about two minutes and deleted after 30 days asleep. The hosted service runs in one region, US East (New York), with no region choice. The documented customer environment option is the desktop app, which runs threads and subagents on the customer's own Mac or PC, and Enterprise offers running in the customer's own cloud on request, with OIDC federation issuing short lived cloud credentials to machines. Sourcedocs.capy.ai/machines, docs.capy.ai/cloud-credentials and capy.ai/enterpriseread 2026-10-28 |
||
|
TemboDeployment & Data Residency In Tembo's cloud each session runs on a dedicated ephemeral Linux VM, and the subprocessor list puts the hosted servers on AWS and Leaseweb in the United States with no region choice. Customers can instead self host the platform as a NixOS image on AWS, Azure or GCP, documented as early access and sales led, and the enterprise page adds bare metal, on premises and air gapped options, though the self hosting docs say updates need outbound HTTPS. Sourcedocs.tembo.io/features/self-hosted/overview, docs.tembo.io/features/sandbox/overview, tembo.io/self-hosted and tembo.io/enterpriseread 2026-10-28 |
||
| Solution readiness | ||
|
Prebuilt Agents, Templates & Packs Ready-made workflows, packaged employees, templates, blueprints, industry solutions, and role-specific agents that reduce time-to-value. |
||
|
CapyPrebuilt Agents, Templates & Packs A plugin marketplace under Settings, MCP and plugins offers searchable, categorized plugins that bundle MCP servers and skills, with a check mark for Verified by Capy and the vendor stating it picks and tests every marketplace plugin; named plugins include Firecrawl, ClickStack and Supermemory, and a team installs them across the organization or per project or adds one from a GitHub repository. An Explore page lists plugins, skills and themes, and built in skills ship with the product beside the Review agent and the Captain experiment. Sourcedocs.capy.ai/integrations/plugins, docs.capy.ai/skills and capy.ai/changelog 0.4.4 and 0.6.0read 2026-10-28 |
||
|
TemboPrebuilt Agents, Templates & Packs Twelve agent templates are published at tembo.io/agent-templates and picked in the app with Use template, and the API lists reusable agent templates with their instructions and required parameters. Ready agents are named by cadence and job, such as daily changelogs, standup summaries, Test Pruner and turning feature requests into Linear tickets. Seven built in skills ship in every sandbox, and repository and managed skills are browsed in a skills marketplace in Settings and through the API. Sourcetembo.io/agent-templates, docs.tembo.io/features/agents and docs.tembo.io/features/agent-skillsread 2026-10-28 |
||
| Platform extensibility | ||
|
Model Flexibility & Routing Ability to work across multiple foundation models, route tasks to different models, or let buyers bring their own providers and keys. |
||
|
CapyModel Flexibility & Routing A person picks the model per thread and per subagent from Anthropic, OpenAI, xAI, Google, Moonshot, Z.ai, DeepSeek, Alibaba, MiniMax, Meta and other makers, can run one task on several models, and can bring enterprise keys for Anthropic, OpenAI, xAI and Google, a gateway or an OpenAI compatible endpoint. Existing Codex, Copilot, SuperGrok and Claude Pro or Max subscriptions connect and draw on the provider's limits, and free models run on the user's own Mac. Sourcedocs.capy.ai/models-and-pricing and capy.ai/changelog 0.4.0 and 0.5.0read 2026-10-28 |
||
|
TemboModel Flexibility & Routing The customer chooses the agent and the model per agent and per session from Anthropic, OpenAI, Google, xAI, Z.ai, Cursor, Fireworks, MiniMax and DeepSeek models on Tembo's gateway, or brings its own keys for OpenAI, Anthropic, xAI, Cursor, Amp, OpenRouter, Bedrock or Vertex, or connects a Claude, ChatGPT or SuperGrok subscription. Tembo ships no model of its own, so model and agent choice stay with the customer. Sourcedocs.tembo.io/features/models, tembo.io/pricing and tembo.io/changelog 2 October 2026read 2026-10-28 |
||
|
APIs, SDKs & MCP Extensibility Composability layer: stable APIs, SDKs, MCP tool consumption/serving, custom tools, and integration into internal systems. |
||
|
CapyAPIs, SDKs & MCP Extensibility Capy documents a public REST API at api.capy.ai with bearer keys and an OpenAPI specification, covering projects, threads, automations, reviews and usage, so developers can start, message, interrupt and archive agent work from their own code. Sourcedocs.capy.ai/api-reference/overviewread 2026-10-25 |
||
|
TemboAPIs, SDKs & MCP Extensibility The Tembo REST API at api.tembo.io, with an OpenAPI spec and keys the customer creates, rotates, expires and revokes at 100 requests a minute, creates sessions, manages repositories and projects, and creates, schedules, triggers and runs agents. A TypeScript SDK (@tembo-io/sdk) wraps it, the tembo CLI pairs the customer's own machine as a device, and Tembo's own MCP server at mcp.tembo.io/mcp connects AI clients to the same API with OAuth or a key. The older tembo.io/mcp page still describes a stdio task management server. Sourcedocs.tembo.io/api, docs.tembo.io/api/mcp and tembo.io/blog 24 September 2026read 2026-10-28 |
||
|
Testing, Debugging & Optimization Testing, debugging, scoring, retries, fallbacks, quality gates, and optimization loops for improving agent workflows before and after deployment. |
||
|
CapyTesting, Debugging & Optimization A built in review agent checks every pull request from a real checkout and labels each finding confirmed or investigate, and agents run the project's own tests and end to end checks in their VMs, since an agent that cannot run the tests cannot verify its work. Those test the customer's code. On the agent side, the same task can run on different models to compare approaches, and there is no evaluation suite or regression testing of agent behavior. Sourcedocs.capy.ai/review and docs.capy.ai/environmentread 2026-10-28 |
||
|
TemboTesting, Debugging & Optimization Tests, lint, type checks and builds run before work returns, a failed workflow auto fix and a Test Pruner template work the customer's own test suite, and agents execute in environments preloaded with the customer's real dependencies rather than a stripped sandbox. Those test the customer's code. There is no evaluation harness, benchmark suite or regression testing for agent behavior, and the GitHub integration docs say self verification is not covered. Sourcetembo.io/review, docs.tembo.io/integrations/github and tembo.io/agent-templatesread 2026-10-28 |
||
| Specialist automation | ||
|
Browser & Computer Use Browser, desktop, or remote/local computer control for workflows that cannot be handled through stable APIs alone. |
||
|
CapyBrowser & Computer Use Each cloud machine runs an XFCE desktop with Google Chrome that streams live into the thread pane, and the agents use native computer use and browser use for end to end testing, recording annotated videos of each run. Cloud iOS and Android phones through Revyl are available from the Pro plan. Capy does not describe approvals for purchases or form submissions. Sourcedocs.capy.ai/machines, docs.capy.ai/cloud-phones and capy.airead 2026-10-28 |
||
|
TemboBrowser & Computer Use Each sandbox ships Chromium 143, VS Code and a desktop with a built in agent-browser skill that automates browser interactions and takes screenshots, and the tembo browser open command. Agents use it to browse sites and operate graphical apps, screen recording captures the run, and a Computer preview streams the desktop at 60 frames a second so a person can watch. Sourcedocs.tembo.io/features/sandbox/built-in-skills-and-tools and tembo.io/blog 15 September 2026 (Tembo Computer)read 2026-10-28 |
||
Pricing snapshot
Sourced from the Index pricing dataset · open each vendor's profile for full detail.
| Pricing | C Capy |
T Tembo |
|---|---|---|
|
Entry price Lowest public entry point |
Lite at $20 a month ($16 billed annually) with $20 of credits. Pro is $100 a month ($80 annual) with $105 of credits, Max runs from $200 to $1,000 a month with credits worth 110% of the price, and Enterprise is custom. | Pro $60 a month and Max $200 a month, each a usage allowance, for up to 5 and 10 users, with priority support on Max. Overage is billed on VM compute per hour and on model tokens. Enterprise is custom, and a free trial covers up to 10 users. |
|
Pricing confidence How public the numbers are |
Public, exact | Public, partial |
|
Billing Primary billing axis |
Monthly or annual subscription tiers that carry an organization wide credit allowance, with no per seat fee. Model tokens and cloud machine time draw the credits down. | A monthly usage allowance per plan, drawn down by VM compute time and model tokens, with overage billing beyond it. |
|
Variable cost Workload / overage exposure |
High variable cost | Medium variable cost |
|
Free tier / trial Try before you buy |
No free tierTrial
|
No free tierTrial
|
|
Buying motion Self-serve vs sales call |
Mixed | Self-serve |
|
Pricing verified When this pricing was last checked |
Product updates
Newest first, each linked to its source.
Capy
-
· Deployment / data residency
The new Dock CLI in Capy 0.7.0 runs Capy without the desktop app, so a build server, cloud VM or spare Mac can host agent threads with the tools and credentials already on it.
Source -
· Workflow orchestration
Capy 0.4.7 lets a Captain, the agent that coordinates a crew of other threads, run on one of the user's paired devices, and the Captain can start crew threads on any paired device or in the cloud.
Source -
· Pricing / packaging
Capy 0.4.0 lets teams run the coding agent on AI subscriptions and API keys they already pay for.
Source
Tembo
-
· Observability / auditability
A Device tab now opens in the session whenever an agent works on one of the user's paired machines, showing that machine's screen live as the agent clicks, types and runs things.
Source -
· Agent capability
When Claude or Codex needs a decision in the middle of a run, it now asks inside the Tembo session, and the user picks one of its options or types an answer so the agent can carry on.
Source -
· Agent capability
Tembo added session forking.
Source
Neither Capy nor Tembo publishes a benchmark or an evaluation harness for agent output, so the quality of a task rests on the model and on review. Tembo documents no spend limit, step limit or per tool permission for its runs, and its approval gate is the customer's own pull request review. Tembo's self hosted platform is in early access. Capy's first public release came in August 2026, and its maker, Scrapybara, Inc., moved from selling hosted computers for agents to building Capy. Tembo began as a managed Postgres company in 2022 and moved to coding agents in May 2025, and its reported funding predates that move. Ask Capy which regions besides US East it can run in, and ask Tembo when self hosting leaves early access.
Other matchups in coding agents
Not the pairing you were after? These compare a different set of coding agents on the same 14 capabilities.