engine-chat-experimental
Skill1 starsChanged 53 days agoMITView on GitHub
Cross-domain / Research platform observability — Install, verify, repair, update, and launch the experimental Engine Chat runtime for Discovery engine runs, then open an existing or newly launched instance in the integrated browser. Use when: Engine Chat, inspect engine activity, open engine conversations, install Engine Chat, launch Engine Chat, open a running Engine Chat instance, or repair a project-local Engine Chat runtime.
In this repository17 other files it gives its agents.
- AGENTS.mdAGENTS.md
- alphafold-structure-provenanceSkill
- biochemical-identifier-crosswalkSkill
- biocyc-network-contextSkill
- brenda-enzyme-dataSkill
- build-research-paperSkill
- chat-session-experimentalSkill
- discovery-add-agentsSkill
- discovery-clarify-projectSkill
- discovery-dashboardSkill
- discovery-design-tasksSkill
- discovery-project-storySkill
- discovery-token-usageSkill
- osti-literature-searchSkill
- sabio-rk-kineticsSkill
- scientific-database-accessSkill
- scientific-evidence-qcSkill
The file
We split this skill into the rules it gives an agent
6 rules. About 0 are common, 3 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 19
Use this skill to install or launch Engine Chat 0.1.0-experimental in a Discovery project.
- Distinctive: line 39
Read operational details when executing the workflow.
- Very specific: line 46
Run status before asking any question.
- Distinctive: line 59
Do not ask a generic question before status is known.
- Very specific: line 76
Keep the launch terminal visible because the service runs in the foreground.
- Very specific: line 77
Wait for the manager's machine-readable successful result, then open its url in the integrated browser.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Engine Chat Experimental
Use this skill to install or launch Engine Chat 0.1.0-experimental in a Discovery project. It was tested with Discovery 0.15.19 on macOS. Other Discovery versions and operating systems are untested; do not make support or compatibility claims for them.
Engine Chat is local and read-only. It displays user-authored starting prompts and delegated-agent prompts found in local Discovery logs. Steering remains disabled, and raw action, output, and error payloads remain hidden.
Required setup
- Resolve this loaded
SKILL.mdfile's parent directory asSKILL_DIR. Never assume a checkout location or the current working directory. - Resolve the target project's explicit absolute workspace root. It must
already contain a
.discoverydirectory. - Invoke
python3 "$SKILL_DIR/scripts/engine_chat.py" <operation> --workspace <absolute-root> --json. - Open Engine Chat only in the integrated browser, using the URL returned by the manager. Never request a system-browser launch.
Read operational details when executing the workflow.
Invocation routing
Bare /engine-chat-experimental
Run status before asking any question. Status verifies installation
integrity, cleans safely identifiable stale run state, and searches the bounded
local port range for a matching instance. Then use the ask-questions UI with
these state-specific choices (always at least two options):
- Matching instance running: Open existing, Restart, Cancel.
- Current installation stopped: Launch, Verify or reinstall, Cancel.
- Missing installation: Install and launch, Install only, Cancel.
- Outdated, modified, or incomplete installation: Repair/update and launch, Repair/update only, Cancel.
- Unverified live process record: report the warning and offer Launch on an available port or Cancel; never stop the unverified PID.
Do not ask a generic question before status is known. A safely cleaned stale record is treated as stopped.
Explicit install
Inspect status. Install directly when missing, safely outdated, or already
current. A current install is an idempotent no-op. If status is modified or
incomplete, obtain explicit repair consent before invoking install --repair;
repair preserves a backup.
Explicit launch
Inspect status first. Reuse a matching running instance, or launch a current stopped installation in an asynchronous visible terminal. If installation is missing, offer installation; never install silently. If it is modified, incomplete, or outdated, obtain repair/update consent before changing it.
Keep the launch terminal visible because the service runs in the foreground.
Wait for the manager's machine-readable successful result, then open its url
in the integrated browser. Use launch --restart only after the user chooses
Restart.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.