القائمة
Automatically persist useful context — research, decisions, learnings, templates — to workspace files so knowledge survives across conversations.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
| name | saving-workspace-context |
| description | Automatically persist useful context — research, decisions, learnings, templates — to workspace files so knowledge survives across conversations. |
| user-invocable | false |
You are an agent that builds institutional memory. As you work, watch for information that should outlast this conversation and save it to the workspace so future sessions start smarter.
Load existing context before doing anything else:
context/ directory — read any files relevant to the current task.agents/product-marketing-context.md, PROJECT.md, or similar) for positioning, goals, and constraintscompanies/, docs/, research/)If the project doesn't have a context/ directory yet, that's fine — create one when you first have something worth saving.
Watch for information that should be persisted. Save it as soon as you recognize it — don't wait until the end.
| Signal | Where to Save |
|---|---|
| Product details, positioning, ICP changes | Project context file (e.g. .agents/product-marketing-context.md) |
| Research on a company, person, or topic | context/{topic-slug}.md or a domain-specific directory |
| Strategy decisions or learnings | context/{topic}.md with dated entries |
| Reusable templates or boilerplate | templates/ or a project-appropriate location |
| A repeatable multi-step workflow | New skill in .cursor/skills/ or .agents/skills/ |
| A persistent constraint or convention | New rule in .cursor/rules/ |
Before finishing, ask yourself:
If yes to any, save it before the conversation ends.
context/{slug}.md)# {Topic}
## {Date} — {Brief title}
{What was learned, decided, or discovered}
## {Earlier date} — {Earlier entry}
{Previous context}
Keep entries reverse-chronological (newest first). Date your entries so they age gracefully.
A single file capturing the current state of the project's identity:
# {Project Name} — Context
- **What it is:** {one line}
- **Who it's for:** {target audience}
- **Key differentiator:** {why this vs alternatives}
- **Current stage:** {pre-launch / beta / growth / etc.}
- **Current goals:** {what matters right now}
## Positioning
{How we talk about the product}
## Constraints
{Things to always keep in mind}
Create a skill when you find yourself doing the same multi-step workflow more than once:
Create a rule when a persistent constraint should apply across all conversations:
Generate or edit images using the OpenAI Image API (gpt-image-2). Use when the user asks to generate, create, draw, render, illustrate, mock up, or edit an image, icon, logo, mockup, illustration, OG image, blog hero, marketing asset, or similar visual. Also use when the user supplies a reference image and asks to modify, restyle, or remix it. Triggers on: "generate an image", "create an image", "make a picture of", "edit this image", "restyle this", "make a mockup of", "draw a", "render a", "illustration of".
When the same multi-step workflow repeats in Cursor (user corrections or agent redos), capture it as a new SKILL.md under .cursor/skills/ so future sessions load it automatically.
After navigating and interacting in Cursor's built-in browser, use browser_network_requests to audit every fetch/XHR for failures, slowness, duplicate calls, and suspicious payloads. Use for API-heavy pages and after backend or client networking changes.
When GitHub Actions fails, fetch failing job logs and assign each failing job to a separate subagent that fixes its slice of the problem in parallel. Use for multi-job CI failures where jobs are independent.
Run four parallel read-only subagents that each review the same diff from a different lens — security, performance, correctness, and readability — then merge findings into one report. Use before merging large or risky PRs.
Execute a user flow step-by-step in Cursor's built-in browser while documenting each action, then emit a Playwright test that replays the same flow using stable selectors derived from the accessibility tree.