sica-fondt/brain/reasoning/momoa/assets/prompts/tool-preambles/screen-capture-tool-preamble.md
Claude 24f816b6a3
Consolidate 22 sibling repos into layered organism structure
Place useful parts of the surrounding repos into sica-fondt by layer, per the
body model (Ada = membrane; brain/endocrine/capabilities/knowledge non-Ada):

- brain/        LLM reasoning + providers (dapr, hermes, MoMoA)
- capabilities/ REPRAG sidecars: hermes tools/skills, dapr tools, parallel
                dispatch, A51 channels, and the OSINT cluster
- knowledge/    LORAG corpus: 754 cyber-skills, agency personas, secure-coding,
                MITRE ATT&CK data
- reference/    defensive threat-reference (C3, shhbruh doc) + AdaYaml parser

License handling: AGPL sources (worldosint, advanced_evolution, mercury,
Reticulum) and GPL DeTTECT are SPEC-only clean-room/port descriptions — no
copyleft code copied. MIT/Apache/data parts copied as working trees.

Safety: shhbruh escape/persistence material and C3 covert-C2 kept as reference
only, not wired into the running organism. See CONSOLIDATION.md.

https://claude.ai/code/session_01UehUqEXXJJCsHoA4voCU5c
2026-06-10 06:53:01 +00:00

25 lines
1.5 KiB
Markdown

---
name: "screen-capture-tool-preamble"
---
Take a screenshot of this project while it's running, providing an image of the Specified Screen. End the session and provide the captured image within the diff / pull request immediately when the image has been created, without waiting for further feedback.
**Screenshot / Screen Capture Instructions**
**Step 1: Install the dependencies**
* pip install shot-scraper && shot-scraper install && npx playwright install && npx playwright install-deps
**Step 2: Start the Project**
* Install any project dependencies.
* Then use `nohup` to start the server in the background.
**CRITICAL:** Your execution environment closes the shell session immediately after a command finishes, so a standard background job will be killed by the hang-up signal (SIGHUP) before I can reach the next step. Use `nohup` to ensure background processes are detached and persist, allowing you to run subsequence commands that require the server to be running to work properly.
**Step 3: Naigate to the correct screen**
* Use playwrite to navigate the app to the correct screen as defined by the user's request.
**Step 4: Capture the Screenshot**
Use the 'shot-scraper' app to take the screenshot. Here's an example of how it works:
shot-scraper http://localhost:3000 -o ./screenshots/screenshot-[timestamp without spaces].png --width 1280 --height 720 --wait 10000
**Specified Screen**
The following instructions specify how the user describe which screen to capture and how to navigate to it:
${UserRequest}