SequentialThinking
PassAudited by ClawScan on May 13, 2026.
Overview
This appears to be a purpose-aligned reasoning-helper plugin, with the main caveats that it can nudge selected models to use its tool and can log generated thoughts locally.
This is reasonable to install if you want a local sequential-thinking helper. Before using it on sensitive tasks, consider disabling thoughtLogging, keeping debug logs off, and leaving the models list empty unless you specifically want the plugin to nudge those models toward using the tool.
Findings (3)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Configured models may choose this tool more often for complex tasks, which can add extra reasoning/tool-call steps.
For configured models, the plugin appends an instruction into the prompt to prefer its own tool. This can affect tool-use decisions, but it is disclosed and scoped to the optional models configuration.
strongly prefer using the `sequential_thinking` tool ... return { appendSystemContext: PREFER_SEQUENTIAL_THINKING_CONTEXT }Leave the models list empty if you do not want prompt nudging, or configure only the specific models where you want this behavior.
Sensitive details included in the model's reasoning steps could appear in local debug logs while the plugin is running.
The tool stores thought inputs in per-run state and logs formatted thought text by default. This is central to the skill, but those thoughts may contain task details or sensitive user-provided information.
this.thoughtLogging = thoughtLogging ?? true; ... state.thoughtHistory.push(input); ... logger.debug(formattedThought
Disable thoughtLogging for sensitive work and avoid putting secrets into prompts or tool thoughts.
If debug logs are enabled, local logs may include session identifiers or tool-call context beyond the visible final answer.
The plugin debug-logs hook context objects for lifecycle/tool-call events. The exact sensitivity depends on what OpenClaw includes in ctx, but full-context logging can expose session or tool-call metadata locally.
logger.debug(`before_tool_call hook triggered, ctx: ${JSON.stringify(ctx)}`);Use this plugin with debug logging disabled for sensitive sessions, and prefer minimal/metadata-only logging in future plugin versions.