- Home
- Skills
- Productivity
- Dayproof Evidence Planner — Mac & MCP Workflows
Works with the AI Agents you already use
Dayproof Evidence Planner — Mac & MCP Workflows
Turn goals into realistic Dayproof plans, avoid duplicates, and check local file evidence with clear human review. Requires the free Mac app and its separate local MCP connection.
Free
Dayproof Evidence Planner — Mac & MCP Workflows
Example session with this skill installed
I have two hours on January 15, 2030 (UTC). Read that day's Dayproof tasks first, then plan work on save/load. My test runner writes reports/2030-01-15/save-load.json with summary.passed as a boolean. Reserve 30 minutes for my own behavior review. Check the report after I run the tests; do not claim you ran them.
- Read your context and instructions
- Compiled the dayproof evidence planner
Illustrative workflow — synthetic validation data
This example demonstrates the packaged developer plan and the real MCP backend checks. It is not a claim about work in your own project.
Saved plan
After checking the existing day, save two tasks for 2030-01-15, UTC:
| Task | Estimated time | Deadline | Completion condition |
|---|---|---|---|
| Implement and test save/load | 90 minutes | 12:00 +00:00 | reports/2030-01-15/save-load.json has summary.passed equal to boolean true |
| Review save/load behavior | 30 minutes | 12:30 +00:00 | You review the agreed scenarios and confirm in Dayproof |
Total estimated time: 2 hours. The report checks one declared result; it does not independently execute tests or assess feature quality. The review remains a manual task.
Evidence results
- Missing report: pending.
- summary.passed is false: pending.
- summary.passed is the string "true": pending, because types must match.
- summary.passed is boolean true: the configured condition passes and this task becomes done.
- Behavior review: still waiting for your confirmation.
Next action
Review the feature against your scenarios, then confirm that task in Dayproof. If the evidence path or acceptance criteria change, update the existing task; that edit clears its previous completion and verdict. Dayproof did not run the tests, send a message, or certify the report's producer.
Connects securely to your tools. The creator never sees your data.
About this skill
Your AI can write a plan. Dayproof keeps it visible on your Mac and checks the evidence you agreed to use.
Dayproof Evidence Planner teaches your agent how to turn goals into a practical working day, save plans through Dayproof's MCP tools, and distinguish passed conditions from work that needs your review. The native Mac app includes a floating task window and desktop cat.
What you get
- Read existing tasks and paginate before creating a plan to avoid duplicates.
- Plan around available time with estimated durations and explicit timezone deadlines.
- Choose file existence, literal text, strict JSON values, Markdown checklists, reading notes, email-draft structure, or manual confirmation.
- Save and read back tasks, check evidence, and preserve plan fields when rescheduling.
- Follow worked developer, course-progress, reading-note, application-draft, carryover, and evening-review workflows.
Includes SKILL.md, three focused reference guides, and three illustrative JSON plans.
Requirements — read before using
This is an optional companion skill. Dayproof itself is free and MIT-licensed at https://github.com/airesearchjishengyin/dayproof . This companion skill is free to use; no purchase is required.
Dayproof v0.2.0 is an early source preview with a Chinese application interface. Build it on your Mac using Git, Node.js 22+, npm, Python 3, and Apple Command Line Tools. The macOS target is 13+; Apple Silicon is the documented tested platform. A signed or notarized installer is not included. The skill instructions are in English.
Your agent needs a local tool host with file access and a separately configured Dayproof STDIO MCP server. Loading this skill through Agensi does not install the app or create that connection. Cloud-only agents cannot reach your Mac through their own localhost.
What evidence means
Background checks run code without LLM calls while Dayproof is running and the Mac is awake. They verify only the configured condition. A checkbox does not prove functioning code, notes do not prove comprehension, and an email draft does not prove delivery. Reading-note and draft checks default to human confirmation. Your AI agent/model has its own usage costs.
Tasks remain in Dayproof's local database. Task details and evidence excerpts returned to an external agent are subject to that provider's data handling. No hosted Dayproof account or paid Dayproof API key is required.
Validation
The three example plans were schema-validated and exercised against the public v0.2.0 backend through real STDIO MCP with isolated synthetic data. Checks covered pagination, missing and strict-type JSON evidence, note and draft structure, human review, and update resets. The public version's 14 automated tests also passed. This validates the backend workflow, not agent-model behavior or every individual agent product.
Source and application documentation: https://github.com/airesearchjishengyin/dayproof .
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Productivity.
Optional companion skill; the free MIT Dayproof app must be built separately. Public v0.2.0 source preview, Chinese app UI, macOS 13+ target; Apple Silicon documented tested. No signed/notarized installer. Requires separate local STDIO MCP access. Text evidence reads are capped at 32 KiB. Structural checks do not prove quality, comprehension, email delivery, report authenticity, or independently executed tests. Real backend/MCP tests passed; individual agent clients were not separately tested.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free to use
No account required to browse
Trust & safety
Security scan
Security scan passed 20 days ago
- Free with an Agensi account
Mac only
- Software
- Dayproof 0.2.0+Dayproof MCP
Git
Node.js 22+npm
Python 3Apple Command Line Tools
Dayproof is a free MIT source preview you build yourself; tested on Apple Silicon. The MCP server runs locally over stdio.
Permissions
Routine operation reads and writes Dayproof tasks through its authenticated loopback MCP bridge and reads only agreed evidence under EVIDENCE_ROOT. Source setup optionally uses a shell, GitHub and npm downloads. DATA_DIR and DAYPROOF_URL select the running local instance; token contents stay local. No public server, email sending, payment, or unrelated account access. Task details/evidence returned to the agent may be processed by its provider.