Productivity & Work · Project Management / Agent Collaboration
TaskFolk
Project-management workspace where human teammates and named AI agents share issues, context and execution state, with agent sessions, permissions, review states, REST API and first-party MCP access.
Free forever: 5 projects, unlimited issues/members, 25 AI credits/month and 128 MB storage. Pro $3/builder/month with unlimited projects, 5,000 AI credits, REST/MCP/skills and 1 GB storage. Business $6/builder/month with 20,000 AI credits and 100 GB.
Taskfolk is a project-management workspace designed so AI agents can participate as first-class members rather than merely appearing in an assistant sidebar. Teams get conventional boards, backlog, lists, calendar, timeline, sprints, milestones and reports, while agents can be assigned issues, leave attributed updates and move work through the same project state as human teammates.12
The agent model is unusually explicit. Assigning work creates a session with states such as pending, running, needs input, review, done, failed and cancelled; stalled sessions can be detected through heartbeats, and completed work without attributed changes can receive an unverified indicator. Per-agent field permissions constrain what an agent can change.2 A first-party MCP server and REST API expose the same workspace operations through scoped keys, while webhooks and audit views support external automation.34
Pricing & access
Taskfolk's Free plan is permanent and includes unlimited members, five projects, unlimited issues, the main project views, GitHub/import tools, 25 AI credits per month and 128 MB storage. Pro costs $3 per builder per month and adds unlimited projects, 5,000 AI credits, REST API, MCP, skills, 1 GB storage and priority support. Business is $6 per builder per month with 20,000 AI credits, 100 GB storage and larger-organization support.5
Viewers are free across plans and AI agents do not count as billed seats. A new workspace also receives a 14-day Pro trial.5 The low builder price is attractive, but teams using AI-heavy features still need to watch the monthly credit allowance rather than treating seat price as the entire cost of use.
Evidence & trust
Taskfolk's documentation is more concrete than its market validation. The public API reference and developer page describe workspace-scoped bearer keys, scoped permissions, rate limiting, audit data and a large REST surface, while MCP re-enters that same authorization pipeline instead of creating a separate uncontrolled path.34 The agent-session design similarly exposes pending/running/review/failure states instead of collapsing all automation into a chat transcript.2
Those are credible architecture facts, but most comparative claims still come from Taskfolk itself. Its pricing and comparison pages position the product against Linear, Jira, ClickUp and others, and its own blog explains why its agent model is distinct; these are useful for understanding the intended differentiation, not independent evidence that Taskfolk performs better.52
Current external coverage does not provide a substantial exact-product verified-user rating corpus, independent uptime/reliability evidence or comparative test of agent assignment, permission enforcement and session-state correctness.
Who it's for
Taskfolk fits teams that already manage work through issues and projects but want coding or other AI agents to participate transparently in that same system. The strongest use case is not generic AI task generation: it is assigning real work to named agents while keeping permissions, attribution, review and project state visible to humans.12
It is a weaker fit for organizations that need a mature enterprise project-management ecosystem with extensive independent market validation, or for teams whose automation is primarily cross-application business workflow rather than issue/project work. AI-heavy teams should also compare the available credit pool with their expected usage.5
Strengths & weaknesses
Strengths
The clearest strengths are low pricing, agent-native project semantics and developer access. Human and AI members share the same issue model, agents receive scoped identities and session states, and REST/MCP integration is documented rather than merely promised.532
The main weakness is evidence maturity. Taskfolk's architecture looks considered, but independent coverage is sparse, the product remains relatively young, and there is little external evidence on support quality, large-workspace performance or how often real agent sessions become stalled, unverified or incorrectly completed. Its value proposition is therefore better established at the design level than at the independently validated outcome level.