feat(gateway): add audit logging for gateway tool calls by HOYALIM · Pull Request #63557 · openclaw/openclaw

HOYALIM · 2026-04-09T05:10:18Z

Summary

Describe the problem and fix in 2–5 bullets:

Problem: gateway-originated tool calls did not have a centralized, append-only audit trail across direct /tools/invoke requests and OpenResponses ingress runs.
Why it matters: operators could not reliably audit tool execution from cron/isolated/gateway-originated paths, and manual log inspection could miss what tool ran with what effective arguments.
What changed: added a dedicated gateway-tool-audit.jsonl helper with tool-argument redaction, threaded gateway audit context through ingress → embedded tool execution, and wired both /tools/invoke and /v1/responses into the same centralized JSONL audit sink.
What did NOT change (scope boundary): no new query UI, rotation policy, HMAC signing, or broader analytics/reporting surface was added in this PR.

Change Type (select all)

Scope (select all touched areas)

Linked Issue/PR

Closes Feature: Gateway-level audit logging for all tool calls #13131
Related [Feature] Session event query API: unlock the event stream agents already produce #17067
This PR fixes a bug or regression

Root Cause (if applicable)

For bug fixes or regressions, explain why this happened, not just what changed. Otherwise write N/A. If the cause is unclear, write Unknown.

Root cause: gateway ingress surfaces had no shared audit sink for tool execution, so direct /tools/invoke and OpenResponses-originated tool calls lacked a centralized append-only record.
Missing detection / guardrail: focused tests covered ingress behavior, but there was no branch-local assertion that gateway-originated tool calls produced a real JSONL audit record with redacted args.
Contributing context (if known): tool execution already flowed through shared wrapper logic, but gateway-only audit requirements were not modeled explicitly in hook context.

Regression Test Plan (if applicable)

For bug fixes or regressions, name the smallest reliable test coverage that should catch this. Otherwise write N/A.

Coverage level that should have caught this:
- Unit test
- Seam / integration test
- End-to-end test
- Existing coverage already sufficient
Target test or file:
- src/gateway/tool-audit.test.ts
- src/gateway/tools-invoke-http.test.ts
- src/gateway/openresponses-http.test.ts
Scenario the test should lock in: gateway-originated tool calls write a centralized JSONL audit record with redacted args, and both ingress surfaces propagate the audit context correctly.
Why this is the smallest reliable guardrail: the behavior is split across one helper and two gateway ingress seams; these tests cover the storage helper, direct /tools/invoke ingress, and OpenResponses ingress without needing broad runtime E2E.
Existing test that already covers this (if any): none for the new JSONL audit sink before this PR.
If no new test is added, why not: N/A

User-visible / Behavior Changes

Gateway-originated tool calls now append structured audit records to ~/.openclaw/logs/gateway-tool-audit.jsonl with redacted tool args and metadata including surface, session, channel, model, runId, and toolCallId.

Diagram (if applicable)

For UI changes or non-trivial logic flows, include a small ASCII diagram reviewers can scan quickly. Otherwise write N/A.

Before:
[gateway ingress] -> [tool execution] -> [no centralized tool-call audit file]

After:
[gateway ingress] -> [gateway audit context] -> [shared hook] -> [gateway-tool-audit.jsonl] -> [tool execution]

Security Impact (required)

New permissions/capabilities? (No)
Secrets/tokens handling changed? (Yes)
New/changed network calls? (No)
Command/tool execution surface changed? (No)
Data access scope changed? (Yes)
If any Yes, explain risk + mitigation:
- Tool-call args are now persisted to a new gateway audit JSONL file.
- Mitigation: args are redacted via existing logging.redactSensitive="tools" behavior before write, and records are written under the existing state/logs path with explicit file-mode restrictions.

Repro + Verification

Environment

OS: macOS (local dev)
Runtime/container: Node 22 + pnpm
Model/provider: OpenResponses model: openclaw; direct /tools/invoke has no model
Integration/channel (if any): gateway /tools/invoke and /v1/responses
Relevant config (redacted): default logging redaction (logging.redactSensitive="tools")

Steps

Trigger a direct gateway /tools/invoke request for an allowed tool.
Trigger an OpenResponses request that reaches gateway-originated tool execution.
Inspect ~/.openclaw/logs/gateway-tool-audit.jsonl.

Expected

Each gateway-originated tool call appends one JSONL record.
Record includes surface, tool, session, channel, model, runId, and toolCallId when available.
Sensitive tool args are redacted before writing.

Actual

Targeted tests and manual JSONL inspection confirmed both ingress paths now propagate audit context, and the emitted JSONL record redacts secret-like values.

Evidence

Attach at least one:

Failing test/log before + passing after
Trace/log snippets
Screenshot/recording
Perf numbers (if relevant)

Trace/log snippets:

{"ts":"2026-04-08T20:00:00.000Z","source":"gateway","event":"tool.call","surface":"tools-invoke","tool":"exec","args":{"command":"OPENAI_API_KEY=sk-sec…7890"},"session":"agent:main:main","channel":"discord","model":null,"runId":"run-1","toolCallId":"call-1"}

Human Verification (required)

What you personally verified (not just CI), and how:

Verified scenarios:
- pnpm test -- src/gateway/tool-audit.test.ts src/gateway/tools-invoke-http.test.ts src/gateway/openresponses-http.test.ts
- manual JSONL append/read check showing redacted tool args in the emitted audit record
Edge cases checked:
- direct /tools/invoke path actually emits an audit record (not just context propagation)
- OpenResponses ingress carries requested model into audit metadata
- secret-like values inside tool args are redacted before file write
What you did not verify:
- full pnpm check / pnpm build as a branch-local clean gate, because latest origin/main still has unrelated pre-existing failures in extensions/msteams and src/agents/skills*

Review Conversations

I replied to or resolved every bot review conversation I addressed in this PR.
I left unresolved only the conversations that still need reviewer or maintainer judgment.

If a bot review conversation is addressed by this PR, resolve that conversation yourself. Do not leave bot review conversation cleanup for maintainers.

Compatibility / Migration

Backward compatible? (Yes)
Config/env changes? (No)
Migration needed? (No)
If yes, exact upgrade steps:

Risks and Mitigations

List only real risks for this PR. Add/remove entries as needed. If none, write None.

Risk: audit file growth over time because this PR adds append-only JSONL writes without rotation.
- Mitigation: scope is intentionally minimal for the initial audit trail; follow-up retention/query work stays separate.
Risk: tool args may contain organization-specific secrets not covered by default patterns.
- Mitigation: existing logging.redactPatterns configuration is honored by the shared redaction helper.

Copilot

Pull request overview

Adds a centralized, append-only gateway audit trail for tool execution that works across both direct POST /tools/invoke and OpenResponses (POST /v1/responses) ingress, with tool-argument redaction and propagation through embedded tool execution.

Changes:

Introduces a gateway-tool-audit.jsonl writer + redaction helper and unit tests.
Wires /tools/invoke to append an audit record per tool call.
Threads gatewayToolAudit context from OpenResponses ingress into embedded runner/tool execution so wrapped tools can emit audit records.

Reviewed changes

Copilot reviewed 12 out of 12 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
src/gateway/tools-invoke-http.ts	Appends a gateway tool-call audit record for direct `/tools/invoke` executions.
src/gateway/tools-invoke-http.test.ts	Mocks/assersts audit append behavior for `/tools/invoke`.
src/gateway/tool-audit.ts	New audit JSONL path resolution, argument sanitization/redaction, and append helper.
src/gateway/tool-audit.test.ts	Verifies JSONL write location and that secrets are redacted.
src/gateway/openresponses-http.ts	Propagates gateway audit context (incl. effective model) into agent command opts.
src/gateway/openresponses-http.test.ts	Asserts OpenResponses ingress sets `gatewayToolAudit` fields.
src/agents/pi-tools.ts	Threads `gatewayToolAudit` option into tool hook context construction.
src/agents/pi-tools.before-tool-call.ts	Emits audit records from the tool wrapper when `gatewayToolAudit` context is present.
src/agents/pi-embedded-runner/run/params.ts	Adds `gatewayToolAudit` to embedded runner params type.
src/agents/pi-embedded-runner/run/attempt.ts	Forwards `gatewayToolAudit` into embedded attempt execution.
src/agents/command/types.ts	Adds `gatewayToolAudit` to `AgentCommandOpts`.
src/agents/command/attempt-execution.ts	Forwards `gatewayToolAudit` from command opts into embedded runner invocation.

greptile-apps · 2026-04-09T05:15:02Z

Greptile Summary

Adds a centralized append-only JSONL audit sink (gateway-tool-audit.jsonl) for gateway-originated tool calls, threading audit context through both the /tools/invoke direct path and the OpenResponses ingress. Redaction is applied via the existing logging.redactSensitive="tools" helper before any write. All remaining findings are P2 style/hardening suggestions and do not block merge.

Confidence Score: 5/5

Safe to merge; all findings are P2 style/hardening suggestions with no correctness or data-loss impact.

Core audit logic (redaction, record shape, path resolution, error swallowing) is correct and consistent across both ingress surfaces. Four P2 findings: file-mode enforcement gap for pre-existing files, non-serializable-args fallback losing structure, duplicated context object in tools-invoke-http.ts, and a misleading test assertion due to a passthrough mock. None affect correctness or produce wrong audit records under normal conditions.

src/gateway/tool-audit.ts (file-mode and args-fallback), src/gateway/tools-invoke-http.ts (context duplication)

Vulnerabilities

Tool args containing secrets are redacted before persisting via the shared redactSensitiveText helper — correct.
The mode: 0o600 on fs.appendFile and mode: 0o700 on fs.mkdir are silently ignored when the file/directory already exists (src/gateway/tool-audit.ts lines 77–78). The stated mitigation ("explicit file-mode restrictions") only holds for the initial creation; a fs.chmod call after append would close the gap.
Blocked/denied tool calls are not emitted to the audit log in either ingress path, which limits post-incident reconstruction of denied attempts.

Prompt To Fix All With AI

This is a comment left during a code review.
Path: src/gateway/tool-audit.ts
Line: 77-78

Comment:
**File mode not enforced on pre-existing audit file**

`fs.appendFile`'s `mode` option and `fs.mkdir`'s `mode` option are both ignored when the target already exists. On first run the file and directory get the intended `0o600`/`0o700` permissions, but on subsequent runs (or if the file was pre-created with wider permissions) those modes are silently skipped. The PR description cites "explicit file-mode restrictions" as a security mitigation, but that guarantee only holds for the initial creation.

Consider adding a `fs.chmod` call after creating the file so permissions are enforced on every run, not just the first:

```suggestion
  await fs.mkdir(path.dirname(auditPath), { recursive: true, mode: 0o700 });
  await fs.appendFile(auditPath, `${JSON.stringify(params.record)}\n`, { encoding: "utf8", mode: 0o600 });
  await fs.chmod(auditPath, 0o600).catch(() => undefined);
```

How can I resolve this? If you propose a fix, please make it concise.

---

This is a comment left during a code review.
Path: src/gateway/tool-audit.ts
Line: 38-45

Comment:
**Fallback for non-serializable args loses all structured data**

When `JSON.stringify(args)` throws (e.g., circular references or `BigInt` values in tool args), the catch branch returns `redactSensitiveText(String(args), ...)`. For any plain object, `String(args)` produces `"[object Object]"`, discarding all arg detail. A sentinel value is more informative:

```suggestion
export function sanitizeGatewayToolAuditArgs(args: unknown): unknown {
  try {
    const raw = JSON.stringify(args ?? null);
    const redacted = redactSensitiveText(raw, { mode: "tools" });
    return JSON.parse(redacted) as unknown;
  } catch {
    // Non-serializable args (circular refs, BigInt, etc.) — return a safe sentinel
    return { _unserializable: true };
  }
}
```

How can I resolve this? If you propose a fix, please make it concise.

---

This is a comment left during a code review.
Path: src/gateway/tools-invoke-http.ts
Line: 292-306

Comment:
**Audit context object duplicated between hook call and record creation**

The same `{ surface, sessionKey, messageChannel, model }` object is built twice — once inside `runBeforeToolCallHook`'s `ctx.gatewayToolAudit` (lines 277–282) and again here inside `createGatewayToolAuditRecord`. If either copy is updated without the other, the hook context and the persisted record drift. Extract to a shared local variable.

How can I resolve this? If you propose a fix, please make it concise.

---

This is a comment left during a code review.
Path: src/gateway/tools-invoke-http.test.ts
Line: 422-432

Comment:
**Test assertion checks a `ctx` field that doesn't exist on the real `GatewayToolAuditRecord`**

`createGatewayToolAuditRecord` is mocked as a passthrough (`vi.fn((params) => params)`), so the "record" seen by `appendGatewayToolAuditRecord` still carries a `ctx` field. In production, `createGatewayToolAuditRecord` flattens `ctx` into `surface`, `session`, `channel`, `model` — no `ctx` key exists on the persisted record. The data-flow check is valid, but consider asserting on the flattened fields or adding a comment explaining the passthrough mock.

How can I resolve this? If you propose a fix, please make it concise.

_{Reviews (1): Last reviewed commit: "fix(gateway): audit direct tools-invoke ..." | Re-trigger Greptile}

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 6d58001930

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".

Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: a7aec1d9e3

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".

Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>

…all-audit-log

clawsweeper · 2026-04-30T04:27:45Z

Codex review: found issues before merge.

Summary
The PR adds a gateway tool-call audit JSONL helper, threads audit context through /tools/invoke and OpenResponses-originated agent runs, and updates focused gateway plus shared guardrail tests.

Reproducibility: yes. The skipped-audit behavior is reproducible by static diff inspection: both proposed append sites are after blocked-call exits, and current main shows those denied branches return before execution.

Next step before merge
Security-sensitive audit behavior and stale current-main routing make maintainer/author branch updates safer than an automated replacement fix PR.

Security
Needs attention: The patch adds a security-sensitive audit log but omits denied gateway tool attempts from that log.

Review findings

[P2] Audit denied direct gateway tool calls — src/gateway/tools-invoke-http.ts:294
[P2] Audit denied embedded gateway tool calls — src/agents/pi-tools.before-tool-call.ts:440
[P3] Add the required changelog entry — CHANGELOG.md:7

Review details

Best possible solution:

Land a gateway-owned audit implementation on the current shared tool invocation and hook paths that records allowed and denied attempts with redacted args, status/reason metadata, focused tests, and a changelog entry.

Do we have a high-confidence way to reproduce the issue?

Yes. The skipped-audit behavior is reproducible by static diff inspection: both proposed append sites are after blocked-call exits, and current main shows those denied branches return before execution.

Is this the best way to solve the issue?

No. The direction is right, but the implementation should audit denied outcomes before returning and should be retargeted to current tools-invoke-shared.ts rather than the older direct HTTP-only path.

Full review comments:

[P2] Audit denied direct gateway tool calls — src/gateway/tools-invoke-http.ts:294
The new /tools/invoke audit append runs only after hookResult.blocked returns. A before-tool hook denial, approval refusal, or loop block currently responds with 403 and no gateway-tool-audit.jsonl row, leaving high-value denied attempts out of the audit trail. Write a redacted denied record before returning.
Confidence: 0.89
[P2] Audit denied embedded gateway tool calls — src/agents/pi-tools.before-tool-call.ts:440
appendGatewayToolAuditEntry is inserted after the wrapped agent tool's blocked branch. OpenResponses-originated tool calls vetoed by plugin policy or approval handling return or throw before the new audit helper runs, so record the denied attempt and reason inside the blocked path.
Confidence: 0.86
[P3] Add the required changelog entry — CHANGELOG.md:7
This PR adds user-facing gateway audit logging and security behavior but does not update CHANGELOG.md. Repo policy requires user-facing feat, fix, and perf changes to have an Unreleased entry.
Confidence: 0.91

Overall correctness: patch is incorrect
Overall confidence: 0.85

Security concerns:

[medium] Denied direct tool attempts are not audited — src/gateway/tools-invoke-http.ts:294
The direct /tools/invoke audit write is placed after the blocked-call return, so policy, approval, or loop denials are missing from the audit trail needed for incident reconstruction.
Confidence: 0.89
[medium] Denied embedded tool attempts are not audited — src/agents/pi-tools.before-tool-call.ts:440
The embedded agent-tool audit write is also after the blocked branch, so OpenResponses-originated denied tool attempts are absent from the proposed JSONL audit log.
Confidence: 0.86

Acceptance criteria:

pnpm test -- src/gateway/tool-audit.test.ts src/gateway/tools-invoke-http.test.ts src/gateway/openresponses-http.test.ts src/agents/pi-tools.before-tool-call.embedded-mode.test.ts
pnpm check:changed

What I checked:

Current main has no audit sink: No current-main matches for gatewayToolAudit, GatewayToolAudit, gateway-tool-audit, or tool-audit under the gateway, agents, tests, docs, or changelog surfaces inspected. (5b2a0fbac198)
Current shared invoke path returns denied calls without auditing: invokeGatewayTool runs runBeforeToolCallHook and returns a tool_call_blocked outcome when the hook blocks, with no audit record creation or append in current main. (src/gateway/tools-invoke-shared.ts:248, 5b2a0fbac198)
PR adds audit helper: The PR diff adds src/gateway/tool-audit.ts with JSONL path resolution, argument redaction, record creation, append, and best-effort chmod handling. (src/gateway/tool-audit.ts:1, d71a2ced0f22)
Direct denied attempts are skipped in PR: In the PR diff, the direct /tools/invoke audit append is after the hookResult.blocked return, so policy, approval, or loop denials exit without a JSONL audit row. (src/gateway/tools-invoke-http.ts:294, d71a2ced0f22)
Embedded denied attempts are skipped in PR: The wrapped agent-tool audit append is inserted only after the blocked-outcome branch, so OpenResponses-originated vetoes or approval denials return or throw before reaching the audit helper. (src/agents/pi-tools.before-tool-call.ts:440, d71a2ced0f22)
Related audit context: The PR is intended to close Feature: Gateway-level audit logging for all tool calls #13131, while feat(gateway): persist tool approval decisions to audit log #72769 is a narrower open follow-up for approval-decision audit logging and does not replace this broader tool-call audit trail.

Likely related people:

steipete: Current-main blame and history point to Peter Steinberger as the dominant maintainer across the gateway tool-invoke, OpenResponses, and before-tool-call hook surfaces, including prior /tools/invoke hardening. (role: recent maintainer; confidence: high; commits: 767fd9f222fd, 06110de6f61e; files: src/gateway/tools-invoke-http.ts, src/gateway/tools-invoke-shared.ts, src/gateway/openresponses-http.ts)
ai-hpc: Authored merged feat(gateway): add SDK-facing tools.invoke RPC #74804, which introduced the current shared tools.invoke RPC/helper path that this PR now needs to target when rebased. (role: adjacent owner; confidence: medium; commits: 6b04521d6a41; files: src/gateway/tools-invoke-shared.ts, src/gateway/server-methods/tools-invoke.ts, packages/sdk/src/client.ts)
vincentkoc: Recent gateway test and regression-harness work touches the same /tools/invoke guardrail area and shared test stability context. (role: recent maintainer; confidence: medium; commits: 688327311c5e; files: src/gateway/tools-invoke-http.test.ts)
Vignesh Natarajan: Introduced the original /tools/invoke HTTP endpoint that the PR is extending with audit logging. (role: introduced behavior; confidence: medium; commits: f1083cd52cf4; files: src/gateway/tools-invoke-http.ts)

Remaining risk / open question:

The branch is stale against current main's shared invokeGatewayTool path, so the audit implementation needs a rebase and likely retargeting before merge.
The required user-facing changelog entry is still missing.

Codex review notes: model gpt-5.5, reasoning high; reviewed against 5b2a0fbac198.

Copilot AI review requested due to automatic review settings April 9, 2026 05:10

openclaw-barnacle Bot added gateway Gateway runtime agents Agent runtime and tooling size: M labels Apr 9, 2026

Copilot started reviewing on behalf of HOYALIM April 9, 2026 05:11 View session

Copilot AI reviewed Apr 9, 2026

View reviewed changes

Comment thread src/gateway/tool-audit.ts

Comment thread src/gateway/tool-audit.ts

Comment thread src/gateway/tools-invoke-http.ts

Comment thread src/agents/pi-tools.before-tool-call.ts

greptile-apps Bot reviewed Apr 9, 2026

View reviewed changes

Comment thread src/gateway/tool-audit.ts Outdated

Comment thread src/gateway/tool-audit.ts

Comment thread src/gateway/tools-invoke-http.ts

Comment thread src/gateway/tools-invoke-http.test.ts

chatgpt-codex-connector Bot reviewed Apr 9, 2026

View reviewed changes

Comment thread src/gateway/tool-audit.ts Outdated

HOYALIM and others added 5 commits April 9, 2026 14:36

feat(gateway): add tool call audit log helper

9411bd2

Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>

feat(agents): thread gateway tool audit context

ae99c35

Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>

feat(gateway): audit ingress tool calls

b2ad8e1

Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>

fix(gateway): audit direct tools-invoke calls

c6df8e7

Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>

fix(gateway): tighten tool audit logging

0cebea1

Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>

HOYALIM force-pushed the feat/gateway-tool-call-audit-log branch from 6d58001 to 0cebea1 Compare April 9, 2026 21:50

fix(gateway): honor custom tool audit redaction

a7aec1d

Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>

chatgpt-codex-connector Bot reviewed Apr 9, 2026

View reviewed changes

Comment thread src/agents/command/attempt-execution.ts Outdated

fix(memory-core): preserve cause on dreams cleanup failure

9fe7312

openclaw-barnacle Bot added the extensions: memory-core Extension: memory-core label Apr 9, 2026

fix(agents): update gateway audit model on fallback

8aedfe7

github-actions Bot mentioned this pull request Apr 10, 2026

🦞 OpenClaw 生态日报 2026-04-10 gsscsd/big_model_radar#163

Open

HOYALIM and others added 3 commits April 9, 2026 18:57

chore(ci): rerun checks

9bd04e7

Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>

Merge remote-tracking branch 'upstream/main' into feat/gateway-tool-c…

1f98de1

…all-audit-log

fix(gateway): preserve compat owner semantics after refresh

6e86d04

openclaw-barnacle Bot removed the extensions: memory-core Extension: memory-core label Apr 11, 2026

HOYALIM mentioned this pull request Apr 13, 2026

test: unblock shared CI regressions on main #65609

Closed

25 tasks

test: fix shared CI regressions

d71a2ce

openclaw-barnacle Bot added the extensions: memory-wiki label Apr 14, 2026

github-actions Bot mentioned this pull request Apr 14, 2026

🦞 OpenClaw 生态日报 2026-04-14 gsscsd/big_model_radar#184

Open

steipete mentioned this pull request Apr 26, 2026

Feature: Gateway-level audit logging for all tool calls #13131

Closed

vincentkoc added the triage: dirty-candidate Candidate: broad unrelated surfaces; may need splitting or cleanup. label Apr 26, 2026

clawsweeper Bot mentioned this pull request Apr 27, 2026

feat(acp): add audit logging system for control plane security #42810

Open

9 tasks

BingqingLyu mentioned this pull request Apr 28, 2026

feat(gateway): add audit logging for gateway tool calls BingqingLyu/openclaw#2198

Open

25 tasks

clawsweeper Bot mentioned this pull request Apr 30, 2026

feat(gateway): persist tool approval decisions to audit log #72769

Open

5 tasks

This was referenced May 1, 2026

feat(gateway): add SDK-facing tools.invoke RPC #74804

Merged

Gateway RPC: add SDK-facing tools.invoke #74705

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

feat(gateway): add audit logging for gateway tool calls#63557

feat(gateway): add audit logging for gateway tool calls#63557
HOYALIM wants to merge 12 commits intoopenclaw:mainfrom
HOYALIM:feat/gateway-tool-call-audit-log

HOYALIM commented Apr 9, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

greptile-apps Bot commented Apr 9, 2026

Vulnerabilities

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

Uh oh!

clawsweeper Bot commented Apr 30, 2026 •

edited

Loading

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Uh oh!

Conversation

HOYALIM commented Apr 9, 2026

Summary

Change Type (select all)

Scope (select all touched areas)

Linked Issue/PR

Root Cause (if applicable)

Regression Test Plan (if applicable)

User-visible / Behavior Changes

Diagram (if applicable)

Security Impact (required)

Repro + Verification

Environment

Steps

Expected

Actual

Evidence

Human Verification (required)

Review Conversations

Compatibility / Migration

Risks and Mitigations

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

greptile-apps Bot commented Apr 9, 2026

Greptile Summary

Confidence Score: 5/5

Vulnerabilities

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

clawsweeper Bot commented Apr 30, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

clawsweeper Bot commented Apr 30, 2026 •

edited

Loading