fix(cli): harden agents list JSON parsing

[WilliamK112]

Summary

• harden agents apply parsing of openclaw agents list --json output
• tolerate leading warning/banner text before the JSON payload
• accept both the current bare array shape and a future { "agents": [...] } wrapper while preserving existing id filtering

Related Issue

Related to #5656 follow-up discussion: the main delete-path bug is covered by #5663; this PR handles the parser hardening that was explicitly left for a separate change.

Verification

• npm test -- src/lib/actions/sandbox/agents/apply.test.ts
• npm run lint -- src/lib/actions/sandbox/agents/apply.ts src/lib/actions/sandbox/agents/apply.test.ts (passes; existing unrelated child-exit-tracker.test.ts unused suppression warning remains)
• npm run build:cli
• npm run typecheck:cli
• git diff --check

Signed-off-by: WilliamK112 89602868+WilliamK112@users.noreply.github.com

Summary by CodeRabbit

• Bug Fixes

  • Improved robustness of agent list retrieval to handle responses with leading text and various JSON formats.

• Tests

  • Added comprehensive test suite verifying agent list parsing handles edge cases and multiple response formats.

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[coderabbitai]

📝 Walkthrough

Walkthrough

Adds two helpers to apply.ts: parseJsonFromOutput scans stdout for the first valid JSON start position, and parseOpenClawAgentsList (exported) accepts a bare array or an { agents: [] } wrapper, filters to entries with a string id, and throws on unexpected shapes. defaultListAgents is updated to call parseOpenClawAgentsList instead of inline JSON.parse. Three new tests cover all supported input formats.

Changes

Defensive OpenClaw agents list JSON parsing

Layer / File(s)	Summary
parseJsonFromOutput and parseOpenClawAgentsList implementation src/lib/actions/sandbox/agents/apply.ts	Adds parseJsonFromOutput to locate the first parseable JSON in mixed stdout and parseOpenClawAgentsList to accept bare arrays or { agents: [] } wrappers, validate id fields, and throw on unexpected shapes. Updates defaultListAgents to call parseOpenClawAgentsList instead of inline JSON.parse.
Tests for parseOpenClawAgentsList src/lib/actions/sandbox/agents/apply.test.ts	Imports parseOpenClawAgentsList and adds a test suite covering a bare JSON array, an object wrapper, and stdout prefixed with leading warning text.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

Poem

🐇 When the CLI warns before its JSON starts,
A careful rabbit scans and finds the [ part,
It strips the noise and plucks the agents free,
No id, no entry — strict as I can be!
Tests confirm three shapes, all passing with cheer,
Defensive parsing, my favorite frontier. 🌟

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title 'fix(cli): harden agents list JSON parsing' directly and accurately summarizes the main change—hardening JSON parsing for the agents list command to handle leading text and multiple JSON shapes.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands.}

[coderabbitai]

🧹 Nitpick comments (1)

src/lib/actions/sandbox/agents/apply.test.ts (1)

103-130: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick win

Add one negative-case assertion to lock the parser’s throw contract.

The positive-path coverage is strong; adding a malformed-shape case will prevent regressions in the explicit “must be array/agents array” behavior.

Suggested test addition

 describe("parseOpenClawAgentsList", () => {
@@
   it("strips leading warning text before the JSON payload", () => {
@@
   });
+
+  it("throws when payload is neither an array nor an object with agents[]", () => {
+    expect(() =>
+      parseOpenClawAgentsList(JSON.stringify({ data: [{ id: "main" }] })),
+    ).toThrow("openclaw agents list --json did not return a JSON array");
+  });
 });

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@src/lib/actions/sandbox/agents/apply.test.ts` around lines 103 - 130, The
parseOpenClawAgentsList test suite currently only covers positive paths (valid
array, object with agents property, and strings with leading warnings). Add a
negative-case test within the describe block that verifies
parseOpenClawAgentsList properly rejects or throws an error when passed
malformed input that is neither a bare array nor an object containing an agents
array property. This will ensure the parser's contract around required input
shapes is locked in and prevents future regressions.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Nitpick comments:
In `@src/lib/actions/sandbox/agents/apply.test.ts`:
- Around line 103-130: The parseOpenClawAgentsList test suite currently only
covers positive paths (valid array, object with agents property, and strings
with leading warnings). Add a negative-case test within the describe block that
verifies parseOpenClawAgentsList properly rejects or throws an error when passed
malformed input that is neither a bare array nor an object containing an agents
array property. This will ensure the parser's contract around required input
shapes is locked in and prevents future regressions.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 5a5c37ab-9734-4f9e-8447-457f643ad03b

📥 Commits

Reviewing files that changed from the base of the PR and between 6571684 and ca9f265.

📒 Files selected for processing (2)

• src/lib/actions/sandbox/agents/apply.test.ts
• src/lib/actions/sandbox/agents/apply.ts

[wscurran]

✨ Thanks for hardening the agents apply parsing to tolerate leading warning text before the JSON payload. This proposes a way to accept both bare array and wrapped object shapes from openclaw agents list --json while preserving existing id filtering.

---

Related open PRs:

• #5663 fix(cli): drop unsupported --non-interactive flag from openclaw agents delete

---

Related open issues:

• #5656 [Linux][Agent&Skills] nemoclaw agents apply does not remove agents absent from manifest

[cjagwani]

Reviewed exact head ca9f265d. The parser keeps the existing empty-output and id-filtering contracts, tolerates bracketed leading warnings by validating complete JSON suffixes, accepts the documented { agents: [...] } wrapper, and still rejects non-array payload shapes. All required checks are green on the current synthetic merge ref; no blocking correctness or security findings.

[cjagwani]

Re-approved 447b8b1 after the signed current-main merge: the reviewed two-file parser diff is unchanged, warning-prefixed bare arrays and object wrappers remain covered, and the full CI suite is green.