agentswitch
Skill2 starsChanged 7 days agoApache-2.0View on GitHub
Check read-only whether an existing Claude Code or opencode QA home can switch to the other harness, including shared-layout, migration-collision, duplicate-registration, MCP-contract, and data-preservation findings. Use for "/agentswitch <claude|opencode>", "Harness wechseln", "switch harness", or before using the other extension. Never migrates, installs, onboards, or runs preflight.
In this repository34 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- accessibility-testingSkill
- api-testingSkill
- bug-reportSkill
- bug-reproductionSkill
- demoSkill
- deploy-to-stageSkill
- exploratory-testingSkill
- failure-analysisSkill
- model-setupSkill
- onboard-productSkill
- onboard-projectSkill
- perf-analysisSkill
- qa-helpSkill
- qa-metricsSkill
- qa-reportSkill
- re-boardSkill
- reconSkill
- release-readinessSkill
- report-to-jiraSkill
- retest-bugSkill
- retro-learnSkill
- risk-based-testingSkill
- scaffold-suiteSkill
- security-testingSkill
- skill-reviewSkill
- test-automationSkill
- test-case-managementSkill
- test-reliabilitySkill
- test-ticketSkill
- visual-testingSkill
- webdriverio-testingSkill
- whats-newSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 9
Check the requested target harness, report the result, and stop.
- Very specific: line 26
Use the session cwd, not an automatically substituted QA-home path.
- Distinctive: line 42
BLOCKED: leave every file and symlink untouched.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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
/agentswitch — read-only harness switch check
Check the requested target harness, report the result, and stop. This skill never performs a migration and never changes the QA home.
Run exactly one check
From the blueprint checkout, with the session directory passed explicitly:
python3 .claude/scripts/agentswitch-check.py <claude|opencode> --cwd <session-directory>In an integrated or shared consumer where the script is linked locally:
python3 .claude/scripts/agentswitch-check.py <claude|opencode> --cwd "$PWD"Use the session cwd, not an automatically substituted QA-home path. The checker
resolves the QA home separately by walking up to .qa-home or the legacy
projects/ + memory/ + evidence/ trio. It never creates the marker or any
missing directory.
Render SOURCE, TARGET, QA home, STATUS, the findings, and the human steps
compactly in the user's language. Preserve the exact status:
READY, READY_WITH_HUMAN_STEPS, or BLOCKED.
What the result means
- READY, shared layout: switching means opening the other extension. Do not suggest installation, migration, scaffolding, or re-onboarding.
- READY_WITH_HUMAN_STEPS: report the old layout and the proposed human
dry-run verbatim when present:
install.sh --mode shared --dry-run. Never run it from this skill. - BLOCKED: leave every file and symlink untouched. Name the collision, dangling/foreign link, duplicate registration, malformed config, or unsafe evidence path. A human resolves it; do not repair it here.
- NO_OP finding: source and target are already the same local harness; do not invent a switch or migration.
Cloud OAuth is a one-time human step per harness. Data Center reuses the existing keychain entry. Trust, provider connection, and model selection remain human steps. The regular preflight belongs to the next test run. Harnesses are used sequentially, never for parallel stage runs.
Read-only boundary
The checker uses only Python's standard library and reads only known config, settings, and symlink paths. It does not write, create directories, invoke subprocesses, access the network or MCP, inspect processes or stage, run an installer or preflight, or access the keychain. Secret files are existence checks only; their contents and environment values are never printed. Project, memory, evidence, and bridge data are likewise existence checks only.
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.

