skills.rest
mmnavarr
mmnavarrCOMMUNITY
Explore all skills by this author →
Website𝕏@mxlcolmGitHub

ce-proof

Run human-in-the-loop markdown review loops via the Proof collaborative editor API.

Updated May 11, 2026
Search Tags:#human-in-the-loop#collaborative editing#proof#markdown review#document feedback#comment threads#tracked suggestions
One-click install
npx skills add https://github.com/mmnavarr/harness --skill ce-proof-mmnavarr
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: ce-proof
Source: https://github.com/mmnavarr/harness/tree/main/skills/ce-proof
Command: npx skills add https://github.com/mmnavarr/harness --skill ce-proof-mmnavarr

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Reviewing markdown drafts (specs, plans, brainstorms) with a human requires clunky copy-paste cycles between chat and documents. This Skill uploads a local markdown file to Proof (proofeditor.ai), lets the user annotate it in a shared web editor, ingests their comments as tracked edits and in-thread replies, and syncs the reviewed doc back to disk. ## Core Features & Use Cases - Shared Document API: Create, read, comment on, and edit shared markdown docs via Proof's Web API with token-based access, baseToken mutation discipline, and idempotency-key retry handling. - HITL Review Loop: A five-phase workflow (upload, ingest feedback, terminal report, next-signal prompt, end-sync) that auto-applies confident fixes as tracked suggestions, escalates genuine disagreements, and atomically writes the final doc back to the local file. - Local Bridge Mode: Drive the macOS Proof app via localhost:9847 for marks, suggestions, presence, and event polling. - Use Case: After drafting a plan with ce-plan, hand the markdown file off for human review — the user highlights text and leaves comments in Proof's web UI, the agent replies in-thread and applies fixes, then syncs the reviewed revision back to disk. ## Quick Start Share the markdown file I just created to Proof so we can iterate on it together with human-in-the-loop review.

Frequently Asked Questions about ce-proof

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I share a markdown file for human review with an AI agent?▼

Upload the markdown to Proof via POST /share/markdown, which returns a shareable tokenUrl. The human annotates in the web UI, and the agent ingests comments, replies in-thread, applies tracked edits, and syncs the final doc back to the local file.

How do I add comments or suggested edits to a Proof document via API?▼

Send ops to POST /api/agent/{slug}/ops with a type field like comment.add or suggestion.add, plus quote, by, and a baseToken from the latest /state read. Use status: accepted on suggestion.add to commit a tracked edit immediately.

Does the Proof API work while a user is editing the document live?▼

Yes. Comment ops, suggestion.add, and block-level /edit/v2 all work during active collaboration. Only rewrite.apply (whole-doc replacement) is blocked with LIVE_CLIENTS_PRESENT when a client is connected.

Why did my Proof API mutation fail with STALE_BASE or BASE_TOKEN_REQUIRED?▼

Every mutation requires a baseToken from the most recent /state or /snapshot read. On STALE_BASE or BASE_TOKEN_REQUIRED, re-read /state and retry once with the fresh token; these are pre-commit errors so retrying is safe.

How do I avoid duplicate comments when retrying failed Proof API calls?▼

Verify before retrying: re-read /state and check whether the intended mark or edit already landed, since timeouts and 5xx responses can occur after the server committed. Include an Idempotency-Key header, reusing the same key only for retries of the identical payload.

Can I edit a Proof doc at the block level instead of quote-based replacement?▼

Yes, use POST /api/agent/{slug}/edit/v2 with an operations array of replace_block, insert_after, delete_block, and similar ops keyed on block refs from /snapshot. Operations commit atomically, making it ideal for bulk mechanical sweeps across many blocks.

Related Skills

Explore More Product & Management Skills→
stefaniukstefaniuk
1

brainstorming

Guides collaborative design exploration and spec writing before any implementation begins.

Community
Advanced
pwretmopwretmo

review-thread-resolution

Resolve GitHub PR review threads via gh api graphql after code changes.

Community
Intermediate
XiaoPuOuOXiaoPuOuO
1

human-in-the-loop-review-ops

Coordinate human-in-the-loop review workflows for AI-generated outputs.

Community
Advanced
liumeixinliumeixin
4

github-code-review

Automate GitHub code reviews with git diff and inline comments.

Community
Advanced