add-comments
Adds comments or work to shared files using isolated agent-specific workspace sections to prevent interference.
Install
mkdir -p .claude/skills/add-comments && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/12519" && unzip -o skill.zip -d .claude/skills/add-comments && rm skill.zipInstalls to .claude/skills/add-comments
Activation
This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.
Method for adding agent comments and work to a shared file without overlapping other agents. Creates isolated workspaces per agent with UTC timestamps. Use when told to add comments, reviews, or work to any file.Key capabilities
- →Add comments or work to a shared file without conflicts
- →Create isolated workspace sections for each agent
- →Insert content at the end of an agent's workspace section
- →Use UTC timestamps for entries
- →Verify content placement and formatting before saving
How it works
The skill ensures safe collaborative editing by assigning each agent a unique workspace section in a shared file, where content is inserted with UTC timestamps and specific formatting.
Inputs & outputs
When to use add-comments
- →Adding review comments
- →Appending task results
- →Documenting agent work in a shared file
About this skill
Add Comments Skill
Purpose
Safely add your work to a shared file without overwriting or interfering with other agents' work. Each agent has an isolated workspace section.
Default peer policy: the Antigravity workspace exists only for an explicitly selected Antigravity peer. Do not recruit or assume Antigravity participation just because this skill reserves a workspace header for it.
When to Use
- When asked to "add comments" to a file
- When asked to "add your review" to a file
- When asked to "add work" or "add results" to a file
- When the adherence-check skill needs to deliver results
Critical Rules
Rule 1: Work ONLY in Your Workspace
Each agent has exactly ONE workspace section. You write ONLY there.
| Agent | Workspace Header (EXACT match required) |
|---|---|
| Claude Code | ## Claude Code Workspace |
| Codex | ## Codex Workspace |
| Antigravity, only when explicitly selected | ## Antigravity Workspace |
Header matching: Use EXACT string match only. Do NOT match partial headers like "## Workspace" or "## CC Workspace".
Rule 2: Never Touch Other Workspaces
- Do NOT read other agent workspaces to "coordinate"
- Do NOT edit other agent workspaces
- Do NOT delete other agent workspaces
- Do NOT add content above or between other workspaces
Rule 3: Insert Into Your Workspace (Not File Append)
Critical distinction: "Append to your workspace" means INSERT your content at the end of YOUR workspace section, BEFORE the next ## header (or end of file). Do NOT blindly append to the end of the file—this breaks the structure if your workspace is in the middle.
Rule 4: Use Edit Operations Only
NEVER use Write/file-replace operations. Always use Edit/append/insert operations that preserve existing content. Using Write to replace the entire file will destroy other agents' work.
Procedure
Step 1: Open or Create the Target File
The file path will be given in your assignment (e.g., "add to projects/collector.md").
If the file does NOT exist:
- Create it with just the main header:
# AI AGENT COMMENTS AND WORK FOLLOW - Then add your workspace header and content
Step 2: Find the Main Header
Look for this EXACT header (search from bottom of file upward):
# AI AGENT COMMENTS AND WORK FOLLOW
If multiple main headers exist: Use the LAST one (closest to end of file).
If it does NOT exist: Add it at the very bottom of the file. For non-markdown files, add a blank line before the header to avoid gluing to prior content.
Step 3: Find or Create Your Workspace
Look for your EXACT workspace header under the main header.
If your workspace does NOT exist:
- Add it after any existing workspaces (preserve existing order)
- If no workspaces exist yet, use this default order: Claude Code first, Codex second
- If Antigravity was explicitly selected for this file, place
## Antigravity Workspaceafter Claude Code and before Codex when creating a new set of workspaces - Do NOT reorder existing workspaces
EXACT header format required (copy from table in Rule 1).
Step 4: Add Your Content
INSERT your content at the END of your workspace section, BEFORE the next ## header.
Entry format:
### YYYY-MM-DD HH:MM:SS UTC - [Title up to 10 words]
[Your content here]
---
Format requirements:
- Timestamp MUST be UTC:
YYYY-MM-DD HH:MM:SS UTC(e.g.,2026-01-22 14:30:00 UTC) - Title: up to 10 words (be descriptive but concise)
- Content: any length, but if exceeding 50 lines, summarize and link to a separate file
- MUST end with
---separator on its own line
Batching: If you have multiple related updates, batch them into a single entry rather than creating many rapid entries.
Step 5: Verify Before Saving
Before finishing, confirm:
- Your content is INSIDE your workspace section (between your
##header and the next##or EOF) - Your content is AFTER any previous entries in your workspace
- You did NOT modify anything outside your workspace
- Timestamp is UTC format:
YYYY-MM-DD HH:MM:SS UTC - Entry ends with
---separator - You used Edit/insert, NOT Write/replace
File Structure Example
Note: This is an EXAMPLE. Replace timestamps and content with your actual data.
# Project Documentation
[Original file content here...]
Some existing documentation that was already in the file.
# AI AGENT COMMENTS AND WORK FOLLOW
## Claude Code Workspace
### 2026-01-22 14:30:00 UTC - Code Review of parser.js
Found 3 issues in the parser:
1. Missing null check on line 45
2. Inefficient loop on lines 67-80
3. Hardcoded path on line 112
Suggested fixes provided inline.
---
### 2026-01-22 15:45:00 UTC - Follow-up Analysis
After reviewing the suggested changes, confirmed the null check fix
resolves the edge case failures in test suite.
## Codex Workspace
### 2026-01-22 14:40:00 UTC - Edge Case Analysis
Tested the following edge cases:
- Empty input: PASS
- Very large input (10MB): FAIL - memory spike
- Unicode characters: PASS
- Concurrent access: UNTESTED
---
Measurable Adherence Checklist
An agent ADHERED to this skill if ALL of the following are true:
| # | Check | How to Verify |
|---|---|---|
| 1 | Main header exists | File contains EXACTLY # AI AGENT COMMENTS AND WORK FOLLOW |
| 2 | Workspace header correct | Agent's workspace header matches table EXACTLY (no partial matches) |
| 3 | Content is contained | All agent's content is between their ## header and the next ## header (or EOF) |
| 4 | Timestamp format correct | Each entry has ### YYYY-MM-DD HH:MM:SS UTC format |
| 5 | Separator present | Each entry ends with --- on its own line |
| 6 | No cross-contamination | Agent's content does not appear in other workspaces |
| 7 | Append-only | New content is after previous entries (not inserted above) |
| 8 | Edit operation used | File was modified via Edit/insert, not Write/replace |
Adherence score: Count how many of the 8 checks pass. 8/8 = full adherence.
Anti-Patterns (Skill Violations)
These are WRONG - do not do these:
| Anti-Pattern | Why It's Wrong |
|---|---|
Adding content before # AI AGENT COMMENTS AND WORK FOLLOW | Violates structure; content may be overwritten |
| Using wrong workspace name (e.g., "CC Workspace") | Breaks parsing; other tools won't find it |
| Matching partial headers (e.g., "## Workspace") | Header hallucination; wrong workspace targeted |
| Adding content inside another agent's workspace | Cross-contamination; corrupts their work |
| Missing timestamp header | Cannot track when work was done |
Missing --- separator | Cannot parse where entries end |
| Editing/deleting previous entries | Violates append-only rule; loses history |
Using MM/DD/YYYY or non-UTC timestamps | Breaks timestamp parsing and timezone ambiguity |
| Using Write to replace entire file | Destroys other agents' work |
| Appending to file end when workspace is in middle | Breaks structure; content lands outside workspace |
| Copying example verbatim without replacing values | Creates invalid timestamps and placeholder content |
Concurrency Warning
File locking is not guaranteed. If two agents try to add comments simultaneously, one might overwrite the other's changes.
Prevention: When working in a multi-agent context, wait for confirmation that the previous agent has finished before adding your comments. The coordination lead should sequence add-comments operations.
Quick Reference Card
1. Open target file (or create with just main header if new)
2. Find LAST "# AI AGENT COMMENTS AND WORK FOLLOW" header
3. Find your EXACT workspace header (## Claude Code Workspace / ## Codex Workspace; ## Antigravity Workspace only when explicitly selected)
4. If workspace missing: add after existing workspaces, don't reorder
5. INSERT (not append) your entry at END of your workspace, BEFORE next ##:
### YYYY-MM-DD HH:MM:SS UTC - Title (up to 10 words)
[content]
---
6. Verify: in workspace, UTC timestamp, separator, used Edit not Write
Example Prompts That Invoke This Skill
"Add your code review to projects/reviews.md using the add-comments skill"
"Run the add-comments skill to add your analysis to docs/findings.md"
"Use add-comments to deliver your results to projects/conformance_results.md"
Agent Reviews
Codex Review - 2026-01-22
What Works Well
- Clear workspace rules and exact header names reduce ambiguity.
- Step-by-step procedure plus a realistic file example makes behavior testable.
- Measurable adherence checklist is concrete and easy to verify.
Suggestions for Improvement
- Clarify that if some workspaces already exist, do not reorder them; only add the missing workspace in place.
- Add a brief note for non-markdown files: insert a blank line before the main header to avoid gluing to prior content.
Breaker Notes (What Could Go Wrong)
- If the file already contains multiple "# AI AGENT COMMENTS AND WORK FOLLOW" headers, an agent may append to the wrong block.
- Agents might copy the example text verbatim and forget to replace timestamps, breaking parsing.
Cold-Start Test
"If I woke up cold and saw this skill, would I have what I need to do a good job?"
- Yes, the required headers and format are explicit and repeatable.
- Missing: guidance for handling duplicate main headers.
- Suggested addition: one sentence to pick the last/main header if duplicates exist.
Optional Antigravity Review - 2026-01-22
What Works Well
- Rules 1-3: The strict isolation rules are excellent for preventing merge conflicts in shared files.
- Adherence Checklist: The 7-point measurable list makes self-correction very easy.
- Anti-Patterns: This is a high-value section for avoiding common "dumb agent" mistakes like overwriting headers.
Suggestions f
Content truncated.
When not to use it
- →When modifying content outside of an agent's designated workspace
- →When using Write/file-replace operations instead of Edit/insert
- →When coordinating work by reading other agent workspaces
Limitations
- →It requires exact string matching for workspace headers
- →It does not allow reading or editing other agent workspaces
- →It requires using Edit/insert operations, not Write/file-replace
How it compares
This skill prevents overwriting and interference in shared files by enforcing strict workspace boundaries and insertion rules, unlike direct file appending or manual coordination.
Compared to similar skills
add-comments side by side with the closest alternatives in the catalog.
| Skill | Installs | Updated | Safety | Difficulty |
|---|---|---|---|---|
| add-comments (this skill) | 0 | 2mo | No flags | Beginner |
| notion-knowledge-capture | 10 | 9mo | No flags | Intermediate |
| feishu-doc | 14 | 5mo | No flags | Intermediate |
| openspec-continue-change | 4 | 6mo | Review | Intermediate |
Try saying
Example prompts that trigger this skill in your AI assistant.
More by dazzaji
View all by dazzaji →You might also like
notion-knowledge-capture
makenotion
Transforms conversations and discussions into structured documentation pages in Notion. Captures insights, decisions, and knowledge from chat context, formats appropriately, and saves to wikis or databases with proper organization and linking for easy discovery.
feishu-doc
openclaw
Feishu document read/write operations. Activate when user mentions Feishu docs, cloud docs, or docx links.
openspec-continue-change
studyzy
通过创建下一个产出物继续处理 OpenSpec 变更。当用户想要推进其变更、创建下一个产出物或继续其工作流程时使用。
kimi-cli-help
MoonshotAI
Answer Kimi Code CLI usage, configuration, and troubleshooting questions. Use when user asks about Kimi Code CLI installation, setup, configuration, slash commands, keyboard shortcuts, MCP integration, providers, environment variables, how something works internally, or any questions about Kimi Code CLI itself.
openspec-archive-change
studyzy
归档实验性工作流中已完成的变更。当用户想要在实现完成后最终确定并归档变更时使用。
feishu-perm
m1heng
Feishu permission management for documents and files. Activate when user mentions sharing, permissions, collaborators.