lvlup-sw/exarchos

test-campaign

Build or improve a repository's native test suite for a stated correctness obligation.

View source

How to use

  1. Copy the installation command and review the source package before installing it in your own agent environment.
  2. Check the source document for required tools, credentials and supported hosts.
  3. Describe your input and desired deliverable; review the output against the source material.
npx skills add https://github.com/lvlup-sw/exarchos --skill test-campaign

Before you start

  • Installation does not include a model subscription, API allowance or permission to access your files.
  • Repository stars and installation counts describe activity, not an independent quality or security test.
Original skill document

Rendered from the source repository. Headings, examples, code, tables, links, and referenced images are preserved.

Test campaign

  1. Pin the repository revision, target configuration, native test command, and accepted behavior. Choose an observable failure that would falsify the behavior. Record any assumption the repository does not settle.
  2. Use test dimensions to choose the checks that can expose that failure. Pair independent checks where either oracle is partial. Keep each test tied to a property, fault, or boundary.
  3. Write tests and replay fixtures in the target's framework. Read a passing test in the same project and match its discovery, assertion, and analyzer conventions. Where a supported testgen command is available, follow generation. Review the predicate, input constraints, transformation, and output projection before using emitted code. Existing vector bindings may retain the legacy contract.
  4. Run the native tests. Verify discovery, observation of the intended branch or fault, and at least one negative control that fails at the intended assertion. Replay concrete failures with their original failure signature. A build error, timeout, or unreached fault leaves the affected claim indeterminate.
  5. Record meaning, oracle, reach, sensitivity, replay, and limits. Include the revision, environment, command, fixture, and pass, fail, or indeterminate result. Keep maintained tests and fixtures in the owning repository and retain raw logs in the existing artifact store.
  6. In a prepared Exarchos task, use the capsule's declared result fields. Return evidence to the existing coordinator. A settlement evidence entry needs a capsule-admitted kind and a reference to an intact recorded evidence row. Retain native reports outside settlement when they have no supported carrier. Keep the product claim distinct from completion of a task that found a defect.

Before execution, write a compact campaign table with the property, selected dimension, accepted oracle, reach condition, negative control, native command, budget, and replay material. Explain omissions that leave a critical path untested. Select methods by the failure they can expose; do not require all 13 dimensions for every change.

Use pass only for the stated property within the recorded run bounds, fail for an observed contract violation, and indeterminate when the evidence cannot decide the claim. Report the test infrastructure result separately: a timeout can violate an accepted liveness deadline, but a harness timeout without a reached target does not establish an application defect. Preserve this distinction through aggregation; do not turn missing evidence into success.

For optional Exarchos composition, inspect the current task capsule and advertised evidence carriers before submitting results. Cite the recorded row, never invent a carrier kind or settlement reference. A supported passing evidence row is not permission to expand its scope. The skill invokes repository commands through the agent's available tools; it adds no scheduler, remote runner, platform abstraction, or second workflow state store.

from this repository

More skills

All skills
lvlup-sw
Community

checkpoint

Save and mutate workflow state — init a workflow, update fields, transition phases, and capture a structured handoff before context exhaustion. Use when the user says 'save progress', 'checkpoint', 'I need to stop', or runs /checkpoint. Writes current workflow phase, task progress, and artifacts to the event store for later resumption. Do NOT use for restoring/reading state after a session break (that is /rehydrate) or for workflow initialization framing handled by ideate/debug/refactor commands

1h installs
1
GitHub stars
19
Updated
Nov 20
lvlup-sw
Community

git-worktrees

Git worktree management for parallel agent team development. Triggers: 'create worktree', 'worktree setup', or during /delegate dispatch. Do NOT use for branch creation without delegation context.

1h installs
1
GitHub stars
19
Updated
Nov 20
lvlup-sw
Community

oneshot

Lightweight workflow for straightforward changes — plan → implement → optional PR. Direct-commit by default; synthesize is opt-in via synthesisPolicy or a runtime requestsynthesize event. Use for trivial fixes, config tweaks, single-file changes, or exploratory work that doesn't warrant subagent dispatch or two-stage review. Triggers: 'oneshot', 'quick fix', 'small change', or oneshot.

1h installs
1
GitHub stars
19
Updated
Nov 20
lvlup-sw
Community

prune

Interactively prune stale non-terminal workflows from the pipeline. Use when the user says 'prune workflows', 'clean stale workflows', 'pipeline cleanup', or runs /prune. Runs a dry-run preview, displays candidates with staleness and safeguard skips, prompts the user to proceed/abort/force, then bulk-cancels approved workflows with a workflow.pruned audit event. Safeguards skip workflows with open PRs or recent commits unless force is set.

1h installs
1
GitHub stars
19
Updated
Nov 20