productive-mcp

An MCP server for Productive.io — log time, inspect projects, manage entries, and get team reports from any MCP-compatible client (Claude Code, Claude Desktop, Cursor, etc.) using plain English:

"log 2.5 hours on the Acme security review project"

"how many hours did Alice log this week?"

"give me my weekly briefing"

Built around the Productive.io JSON:API v2 with fuzzy project and person matching, a local disk cache, billing-cutoff-aware period resolution, and per-project default-service memory.

Features

• 14 tools covering projects, people, services, time entries, reports, and briefings

• Cross-platform credential storage — macOS Keychain, Windows Credential Manager (DPAPI/TPM-backed), Linux Secret Service, or environment variables

• Fuzzy matching for projects and people — "Acme", "1099 Acme", "Alice", or "me" all resolve naturally

• Billing-cutoff-aware period resolution — "this_month" shifts based on your invoicing cycle

• Remembers your default service per project so you don't have to specify it every time

• Local cache for projects, people, and services (1-hour TTL, manually refreshable)

• TOON output mode — optional token-optimised encoding for 30–60% fewer tokens

• hours in, hours out — the API uses minutes internally, but you never see them

• Scoped to "me" by default — list_time_entries only shows your own entries unless you opt out

Requirements

• Python 3.11+

• A Productive.io account with API access enabled

• macOS, Windows, or Linux

Install

1. Clone and install

git clone https://github.com/cameronfairbairn/productive-mcp.git
cd productive-mcp
uv venv
uv pip install -e .

(or python -m venv .venv && .venv/bin/pip install -e . if you don't use uv)

2. Get your Productive credentials

You need three values:

3. Store credentials

Option A — OS credential store (recommended)

Works on any platform via the keyring library:

python -c "import keyring; keyring.set_password('productive-mcp', 'token', '<token>')"
python -c "import keyring; keyring.set_password('productive-mcp', 'org_id', '<org_id>')"
python -c "import keyring; keyring.set_password('productive-mcp', 'person_id', '<person_id>')"

On macOS, the security CLI also works (keyring reads from the same Keychain):

security add-generic-password -s productive-mcp -a token      -w "<token>"      -U
security add-generic-password -s productive-mcp -a org_id     -w "<org_id>"     -U
security add-generic-password -s productive-mcp -a person_id  -w "<person_id>"  -U

Option B — environment variables (CI / headless / override)

export PRODUCTIVE_MCP_TOKEN="<token>"
export PRODUCTIVE_MCP_ORG_ID="<org_id>"
export PRODUCTIVE_MCP_PERSON_ID="<person_id>"

Environment variables take precedence over the credential store.

4. Register the server with your MCP client

Claude Code (~/.claude.json)

{
  "mcpServers": {
    "productive": {
      "type": "stdio",
      "command": "/path/to/productive-mcp/.venv/bin/productive-mcp",
      "args": []
    }
  }
}

Claude Desktop (claude_desktop_config.json)

{
  "mcpServers": {
    "productive": {
      "command": "/path/to/productive-mcp/.venv/bin/productive-mcp"
    }
  }
}

Restart the client after editing its config.

5. (Optional) Global install with the bundled deploy script

bash scripts/install.sh

This creates ~/.local/share/productive-mcp/ containing a fresh venv and a run.sh launcher. Point your MCP client at ~/.local/share/productive-mcp/run.sh instead. Re-running the script upgrades in place.

Tools

All tools are prefixed with productive_ so they namespace cleanly alongside other MCP servers.

Time tracking

Reports and briefings

Projects, people, and services

Administration

Key tool details

productive_get_time_report

period   : str?  — "this_month", "last_month", "this_week", "last_week"
after    : str?  — ISO date; ignored if period set
before   : str?  — ISO date; ignored if period set
person   : str?  — Name, email, id, or "me" (default)
project  : str?  — Scope to a specific project

Returns totals_hours: {worked, client, internal, holidays}. When PRODUCTIVE_BILLING_CUTOFF_DAY is set, "this_month" shifts based on your billing cycle.

productive_my_briefing

period : str — Default "this_week"

Returns total hours, projects touched, hours-by-project breakdown, and recent entries. Designed for the Monday-morning "what am I working on?" question.

productive_get_employee_hours

person : str  — Fuzzy name, email, or "me"
period : str? — Symbolic period (default: no filter)
after  : str? — ISO date
before : str? — ISO date

Finds a person by fuzzy match, returns their hours for the period.

Configuration

Environment variables

Billing cutoff day

The PRODUCTIVE_BILLING_CUTOFF_DAY controls how symbolic periods resolve for invoicing workflows:

• Before cutoff day: "this_month" = previous calendar month (you're still closing invoices)

• From cutoff day onward: "this_month" = current calendar month

• Unset: "this_month" always means the current calendar month

Example: with PRODUCTIVE_BILLING_CUTOFF_DAY=10, on April 5th, "this_month" resolves to March 1–31.

Security & permissions

Credential storage

API tokens are never written to disk by this project. They live in the OS credential store (Keychain / Credential Manager / Secret Service) or in environment variables.

Productive's permission model

Each team member generates their own API token via Productive → Settings → API integrations (docs). Tokens inherit the generating user's permissions:

"API tokens inherit the same access restrictions as the user they are associated with. If a user does not have permission to access certain features or data within Productive, these limitations will also apply to their API token."

Tokens also have a read-only vs read/write scope chosen at creation time.

What this means for team deployments:

• A "Member"-role user's token cannot query other members' time entries — Productive returns 403.

• The real access boundary is the Productive role (Settings → People → Role), not MCP-side configuration.

• Tools like productive_get_employee_hours will simply return empty/error for users whose Productive role doesn't grant team visibility.

For more on roles: User Permissions Overview | Productive API docs

How it works

Architecture

┌─────────────────────┐   stdio    ┌──────────────────────┐   HTTPS   ┌─────────────────┐
│  MCP client         │ ─────────► │  productive-mcp      │ ────────► │  Productive.io  │
│  (Claude Code etc.) │            │  (FastMCP server)    │           │  JSON:API v2    │
└─────────────────────┘            └──────────┬───────────┘           └─────────────────┘
                                              │
                                              ▼
                                    ~/.config/productive-mcp/
                                    ├── cache.json        (projects, people, services)
                                    └── preferences.json  (per-project default service)

Credential lookup order

1. Environment variable (PRODUCTIVE_MCP_TOKEN / _ORG_ID / _PERSON_ID)

2. OS credential store via keyring (Keychain / Credential Manager / Secret Service)

3. Error with a helpful message pointing at both options

Local state

Two files under ~/.config/productive-mcp/, both 0600:

• cache.json — projects, people (active only), per-project services. 1-hour TTL.

• preferences.json — { "default_services": { "<project_id>": "<service_id>" } }.

Neither file ever contains credentials.

Development

uv pip install -e ".[dev]"

# unit tests (no network required)
pytest

# integration tests (hit the real Productive API — requires credentials)
pytest -m integration

# lint + type check
ruff check .
mypy src

Project layout

src/productive_mcp/
├── __main__.py      Entrypoint (python -m productive_mcp)
├── server.py        FastMCP tool definitions (14 tools)
├── client.py        Async Productive.io API client + fuzzy matcher
├── auth.py          Cross-platform credential loader (keyring + env)
├── storage.py       Local cache + preferences persistence
├── periods.py       Billing-cutoff-aware period resolution
└── reporting.py     Time entry categorisation for reports
scripts/
└── install.sh       One-shot deploy script for the global launcher
tests/
├── test_client.py   Fuzzy matching + trim functions
├── test_periods.py  Period resolution + billing cutoff
├── test_reporting.py  Entry categorisation
└── test_formatter.py  TOON output encoding

Troubleshooting

"Credential store lookup failed" The keyring backend isn't configured or accessible. Check: python -c "import keyring; print(keyring.get_keyring())". On headless Linux, use environment variables instead.

"No credential found for service=productive-mcp" Credentials haven't been stored yet. Run the keyring.set_password(...) commands from the install section.

"No project matches 'X'" Cache may be stale. Call productive_refresh_cache and retry.

"Ambiguous project query" Two projects scored nearly identically. Be more specific — add the project number.

"Project has multiple services; pass service_hint" Either pass service_hint="…" (it'll be remembered), or call productive_set_default_service once upfront.

Alternatives

Several other Productive.io MCP servers exist:

• berwickgeek/productive-mcp — the most feature-complete general-purpose server (projects, tasks, boards, people, workflows). Node.js. No time tracking.

• adamchrabaszcz/productive-time-mcp — companion time-tracking server for the above.

• druellan/Productive-Simple-MCP — read-only Python/FastMCP server using TOON output for low token usage.

• laurkee/productive-mcp (Codeberg) — ticketing-focused.

Why pick this one?

1. Cross-platform credential store — not a .env file. macOS Keychain, Windows Credential Manager (TPM-backed), Linux Secret Service.

2. Fuzzy matching for projects and people — say "Alice" or "1099 Acme" instead of looking up IDs.

3. Per-project default service memory — log time without specifying the service after the first call.

4. Billing-cutoff-aware reports — "this_month" means what your invoicing cycle says it means.

5. TOON output mode — 30–60% fewer tokens for long sessions.

License

MIT — see LICENSE.

Contributing

Issues and PRs welcome. Please include tests for new behaviour.

Not affiliated with Productive.io. "Productive" is a trademark of its respective owners.