chore: prune verified-dead core code (cli, sdk, resources, scripts)#409
chore: prune verified-dead core code (cli, sdk, resources, scripts)#409EfeDurmaz16 wants to merge 5 commits into
Conversation
demo.py: drop show_post_payment_experience and the five degraded-state handlers (handle_wallet_creation_failure / faucet_empty / port_in_use / rpc_down / no_network) — defined but called nowhere (~135 lines). api.py: drop get_client(), which referenced click.ClickException without importing click (would NameError if ever called) and is imported nowhere.
buildUrl allocated `const url = new URL(path, baseURL + '/')`, never used it (the real URL is built into `u` via string concat), then suppressed the unused-var warning with `void url;`. Remove both. tsc --noEmit clean.
…sury list_financial_accounts (async + sync) initialised params["refresh"] with a bool and overwrote it on the next line with str(refresh).lower(). Set it once. Behaviour unchanged (final value identical).
The second `if status == COMPLETED and conclusion in {SUCCESS, SKIPPED}`
was unreachable — the preceding check already `continue`s for any
conclusion in {SUCCESS, SKIPPED}. Removed it and the now-unused `status`.
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 480fd0e1b9
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
|
|
||
| 1. Create or update a budget with `sardis_set_budget` before the first spend. | ||
| 2. Run `sardis_check_policy` before every `sardis_spend`. | ||
| 3. Continue only when the policy result is `allow`. |
There was a problem hiding this comment.
The skill pack advertises that these instructions work with the Sardis MCP server (packages/sardis-openclaw/README.md lines 17-19), but this flow only permits spending when sardis_check_policy returns the literal agent-tools outcome allow. The MCP implementation serializes policy results as allowed: true/false and decision.outcome: 'APPROVED' | 'BLOCKED' (packages/sardis-mcp-server/src/tools/policy.ts lines 64 and 412), so an OpenClaw agent using the supported MCP server will fail closed even for allowed payments because step 3 never matches. Include the MCP result shape, or scope the skill to agent-tools outcomes, before requiring the spend step.
Useful? React with 👍 / 👎.
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 480fd0e1b9
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
|
|
||
| 1. Create or update a budget with `sardis_set_budget` before the first spend. | ||
| 2. Run `sardis_check_policy` before every `sardis_spend`. | ||
| 3. Continue only when the policy result is `allow`. |
There was a problem hiding this comment.
The skill pack advertises that these instructions work with the Sardis MCP server (packages/sardis-openclaw/README.md lines 17-19), but this flow only permits spending when sardis_check_policy returns the literal agent-tools outcome allow. The MCP implementation serializes policy results as allowed: true/false and decision.outcome: 'APPROVED' | 'BLOCKED' (packages/sardis-mcp-server/src/tools/policy.ts lines 64 and 412), so an OpenClaw agent using the supported MCP server will fail closed even for allowed payments because step 3 never matches. Include the MCP result shape, or scope the skill to agent-tools outcomes, before requiring the spend step.
Useful? React with 👍 / 👎.
1 participant
Four atomic, behaviour-preserving removals surfaced by the minimization audit:
demo.pyshow_post_payment_experience + 5 degraded-state handlers (called nowhere, ~135 lines); dropapi.pyget_client (referenced click without importing it — would NameError; imported nowhere).URLallocation +void url;inengine.buildUrl.params['refresh']in treasury list_financial_accounts (async+sync).non_success_check_count.Verified:
py_compile+tsc --noEmit(0 errors).