Test Slack authentication, message creation and cleanup, and OpenClaw message handling. Collect evidence from stored messages and test runs, while distinguishing what API checks prove from interactions that require observation in the actual Slack client.
Test Slack with Convex-leased user OAuth or QA bots; check credentials, exercise owned messages, prove Gateway behavior, and distinguish API evidence from real Slack client interactions.
slack-e2e tests Slack authentication, message operations, and OpenClaw message reception and replies while keeping records. It checks user authentication and tests creating, reading, editing, and deleting test messages. Within the existing QA Lab, it verifies replies through the connecting Gateway, file operations, and restarts. It also checks stored results and cleanup.
Use it to check a Slack integration or identify missing permissions. It helps you report stored-message checks separately from actual client interactions, rather than judging success from API responses alone. The resulting evidence and cleanup checks support a report of what was verified.
OpenClaw message reception and replies, then summarize the evidence and cleanup results.”Requires a source environment with dependencies installed, an existing Convex login, available credentials, and Slack permissions. Test posts may be visible to existing listeners. Checking clicks, slash commands, or rendering requires an authorized, logged-in Slack client. API records alone cannot prove client interactions. Credentials and raw records must remain private.
This overview was written by AI (OpenAI Codex) from the public SKILL.md. Check the original below for the exact behavior.
AI rated this skill on four points using only the contents of its SKILL.md. Stars and popularity are not taken into account.
The total counts each point as A = 25, B = 15, C = 5. Rated by AI (OpenAI Codex) as of Oct 19, 2026. Treat it as a guide and read the original before installing.
Review the instructions the agent will receive before you install.
| Goal | Entry point | Boundary |
|---|---|---|
| Check shared user OAuth | Read-only command below | User/workspace identity and channel history; no Gateway |
| Exercise user-authored text | Same command with --smoke | Create, read, edit, delete, verify absence; no Gateway reply claim |
| Prove OpenClaw ingress, replies, tools, files, or restart | Existing QA Lab lane below | Leased driver/SUT bots and an isolated Gateway |
| Prove real clicks, slash commands, or rendering | Authorized Slack client | Actual client observation, not reconstructed API data |
Start from a source checkout with normal dependencies and an existing Convex
login. If needed, ask the operator to run convex login or the cached
bunx --no-install convex login. The broker provides the authorized pool;
agents do not need to copy Slack tokens. Login alone cannot supply a missing
pool credential or grant Slack permissions.
node .agents/skills/slack-e2e/scripts/user-oauth-smoke.mjs
node .agents/skills/slack-e2e/scripts/user-oauth-smoke.mjs --smoke
The default is read-only in Slack; both modes acquire, heartbeat, and release
a Convex lease. --smoke creates one synthetic, unmentioned top-level message,
checks stored authorship, edits it, and deletes that exact message. Existing
workspace listeners can still observe it. Each command uses a new private
artifact directory and prints a safe result plus its path.
Read user OAuth setup and recovery for the pool payload,
permissions, result interpretation, and failure handling. This skill-owned
command reuses the existing lease helper; it changes no QA Lab code or release
lane and starts no Gateway or model. It does not add a user-driver option to
qa slack.
Use the source-checkout QA Lab owner for Gateway proof, not a separate bot runner. A run owns a leased driver/SUT pair, a temporary Gateway, native fixture receipts, and private artifacts. The driver has no Socket Mode connection: the SUT Gateway exclusively owns the app token and event delivery.
Use the existing Convex login above. No Slack tokens, broker secrets, model credentials, app creation, or scope grants belong in the command line.
Inspect the available scenarios and run readiness:
pnpm openclaw qa slack --list-scenarios
pnpm openclaw qa slack --doctor --output-dir .artifacts/qa-e2e/slack-doctor
Doctor checks the actual lease: distinct bot identities in one workspace, both bots' channel access, the owned Gateway connection, and advertised OAuth scopes. Known missing scopes fail with their actor and exact names. Missing scope metadata means mutation capability is unverified, not granted. Missing optional file/reaction scopes do not block text-only flows; they block the affected native operations before dispatch. Doctor's full lifecycle capability verdict remains separate from mandatory connection readiness. A pool/login/permission failure is an owner prerequisite, not a reason to rotate credentials until one passes. Every subsequent run checks its own lease.
Run the native lifecycle recipe:
pnpm openclaw qa slack \
--scenario-file qa/scenarios/channels/slack-e2e-lifecycle.yaml \
--output-dir .artifacts/qa-e2e/slack-lifecycle
--doctor and --scenario-file default to Convex, the CI role, and
mock-openai. This is real Slack/Gateway transport with a deterministic model.
Explicit overrides remain available; ordinary curated qa slack defaults do
not change. Use a unique output directory per run. Additional
--scenario-file arguments select additional complete YAML scenarios.
Read the summary, evidence, and cleanup result before claiming success. Choose feature recipes for native operations or model-tool evidence, and runtime recipes for config/restart, custom flows, cancellation, artifacts, and recovery.
| Surface | What automation proves |
|---|---|
| Mention/quiet ingress, owned edits/deletes, reactions, upload, thread replies | Fixture/API operations and explicit stored readback; not model tool calls |
| Correlated SUT reply | A stored reply from the leased SUT after ingress; mock or live model must be named |
| Gateway debug-proxy capture | Slack accepted a Gateway API write; neither Socket Mode event delivery nor rendering |
| Block Kit, file IDs, reaction state | Stored structure/identity only |
| Human slash invocation, real button clicks, Agent View, visible rendering, human typing | Manual/Mantis client lane; bot/API evidence is insufficient |
For visual or human-interaction requests, use an authorized, logged-in Slack
client; the maintained Mantis workflow is in
docs/concepts/qa-e2e-automation/operator-flow.md. Check which surface it actually
captured: --approval-checkpoints renders API message data, not the Slack client.
A Slack sign-in screenshot is not proof of a workspace action.
Preserve inspected client screenshots alongside native evidence. Posting slash
text through the API is not a slash invocation; Gateway approval RPCs are not
button clicks. Do not infer typing from assistant status or start another Socket
Mode recorder with the SUT app token.
Report scenario/model lane, stored native identities or safe counts, artifact paths, cleanup outcome, missing permissions, and any manual-only gap. Keep raw Gateway capture, channel IDs, messages, media, and lease material private; publish only sanitized proof. A successful API response with failed stored readback, an uncertain send, or failed cleanup is not an end-to-end pass.
No reviews yet. Tell others what you thought.
No login needed. Share how it worked for you, ask a question, or note where it did not run.
No comments yet. Be the first to write one.
Set up and authenticate the 1Password CLI, then supply stored passwords and API keys to commands or configuration. The skill covers desktop integration, service accounts, and interactive sign-in while keeping secrets out of chat and logs.
Route requests from OpenClaw to external coding agents such as Claude Code or Cursor. Start or continue sessions using ACP, a protocol for connecting agents, or relay prompts directly through acpx while preserving conversation context.
Add a model provider to OpenClaw and verify it with a real response. The skill validates configuration, keeps credentials in environment or file references, and reports test results and any default model changes.
Prepare an agent activity transcript for a GitHub pull request or issue. Extract relevant prompts, decisions, and results from local logs, remove sensitive material, and preview the edited record before separately authorized publication.
Manage Apple Notes on macOS through memo, a command-line tool. Create, browse, search, edit, delete, and move notes between folders, or export selected notes to HTML or Markdown for use outside the app.
Manage Apple Reminders from your Mac with remindctl, a terminal tool. View tasks by date, add or edit reminders, organize lists, and mark tasks complete or delete them, including personal to-dos that sync to iOS.