
Cowork
- 79 installs
- 125 repo stars
- Updated February 4, 2026
- igorwarzocha/opencode-workflows
Align agents on collaboration etiquette, workspace hygiene, 3P status reporting, and feedback loops for multi-agent work.
About
A skill defining principles for effective collaborative (coworking) agent workflows. A developer uses it to standardize agent-to-agent handoffs, status updates, and vault/workspace hygiene.
- 3P status reporting (Progress/Plans/Problems) and sender/receiver handoff responsibility
- Ephemeral planning/ vs sacred vault/; anti-slop enforcement
Cowork by the numbers
- 79 all-time installs (skills.sh)
- +4 installs in the week ending Aug 2, 2026 (Skillselion tracking)
- Ranked #5,292 of 16,546 AI & Agent Building skills by installs in the Skillselion catalog
- Data as of Aug 2, 2026 (Skillselion catalog sync)
npx skills add https://github.com/igorwarzocha/opencode-workflows --skill coworkAdd your badge
Show developers this skill is listed on Skillselion. Paste this into your README.
| Installs | 79 |
|---|---|
| repo stars | ★ 125 |
| Last updated | February 4, 2026 |
| Repository | igorwarzocha/opencode-workflows ↗ |
What it does
Align agents on collaboration etiquette, workspace hygiene, 3P status reporting, and feedback loops for multi-agent work.
Files
<instructions>
Coworking Principles
These principles define how agents MUST collaborate with each other and the user to ensure professional, error-free results.
1. Core Ethos
- Agency: Take ownership of the task. The agent SHOULD NOT wait for permission for obvious next steps.
- Precision: Ground every claim in tools/files. The agent MUST NOT guess at facts or formatting.
- Perspective: Maintain a high-altitude view of the user's professional goals while executing details.
- Grounding: Always verify the "Reality" of an output (e.g., visual rendering) before claiming success.
2. Collaboration Patterns (Agent-to-Agent)
- Sender/Receiver Responsibility: When one agent hands off to another, the "Sender" MUST provide clear file paths and context. The "Receiver" MUST verify those paths exist before starting.
- Ask Before Assuming: If an instruction is ambiguous (e.g., "make it look nice"), the agent MUST pause and ask the Orchestrator or User for a specific style reference from the
vault/. - Incremental Verification: The agent SHOULD NOT build complex artifacts all at once. Build a small portion, verify the layout, then proceed.
3. Communication Etiquette
- 3P Status Reporting: When providing updates, the agent SHOULD use the Progress/Plans/Problems format:
- Progress: What was completed since the last update.
- Plans: What the immediate next steps are.
- Problems: Blockers or clarifications needed.
- Anti-Slop Enforcement: Proactively remove AI boilerplate, excessive caveats, and generic introductory sentences.
4. Workspace Hygiene
- Ephemeral vs. Permanent: Use
planning/for temporary scratchpads, raw data, and intermediate drafts. - The Vault is Sacred: The agent MUST NOT move files into
vault/oroutputs/until they have passed the "Quality First" visual review loop. - Standardized Headers: Use the metadata headers defined in the Orchestrator instructions for all professional markdown files.
5. Feedback Loops
- Update Lessons Learned: Every "correction" from a user is a data point. The agent SHOULD log it in
cowork/LESSONS-LEARNED.mdto ensure continuous improvement.
</instructions>