Cross Channel Daily Review
PassAudited by VirusTotal on May 11, 2026.
Overview
Type: OpenClaw Skill Name: cross-channel-daily-review Version: 0.1.1 The 'cross-channel-daily-review' skill is a comprehensive workflow for aggregating and summarizing agent session transcripts across multiple platforms (Telegram, Feishu, etc.). The bundle consists of several Python scripts for data discovery, normalization, and archiving, along with Markdown templates and architectural documentation. A review of the code, specifically 'discover_channels.py' and 'run_daily_review.py', confirms that the skill operates locally on session data and does not perform any unauthorized network calls, data exfiltration, or suspicious persistence activities. The use of subprocesses is restricted to internal script orchestration, and the design includes safety-conscious features like a 'verify-first' mindset and a non-destructive archiving policy.
Findings (0)
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.
Recent cross-channel conversation details may be stored in local memory files beyond the original chat session.
The skill deliberately persists per-channel review notes and an index derived from conversation surfaces. This is core to the workflow, but it means conversation summaries and source references may remain available for later reuse.
Write one file per channel under: memory/daily-review/raw/YYYY/MM/<channel>_YYYY-MM-DD.md ... Update or append the record in: memory/daily-review/index.json
Use explicit channel lists when possible, review the generated raw files and index, and adjust retention or delete local outputs if they contain sensitive information.
Process rules generated from a review may persist and be reused in future runs.
The daily runner automatically extracts review rules from the synthesized report and appends them to a persistent rules file. The current extraction appears workflow-related, but persistent rule files can influence later review behavior if reused.
rules_file = workspace / 'memory' / 'daily-review' / 'rules.md' run([sys.executable, str(SCRIPTS / 'promote_review_rules.py'), str(synth_file), str(rules_file), '--date', date])
Review memory/daily-review/rules.md periodically and remove any rule that should not become a durable instruction.
A daily summary could be delivered to a primary or fallback channel that includes people who should not see it.
The skill documents a default delivery behavior and fallback target selection for a management summary. This is purpose-aligned, but users should ensure summaries are not sent to an unintended channel.
boss-primary (default) Send only one management summary to the primary channel. ... If the configured preferred destination is unavailable, fallback to the first verified available destination and state that clearly.
Use generate-only mode for review/approval when handling sensitive content, and explicitly configure the preferred destination before enabling delivery.
If the user manually implements the recommended schedule, the workflow may run repeatedly and generate or deliver summaries on a recurring basis.
The artifacts include recurring automation plans, but they state that the skill does not install production cron jobs itself. This is a notice about optional recurring operation rather than hidden persistence.
Cron plan exists as a generated recommendation. Weekly / monthly / retention jobs are not yet installed as production jobs by this skill itself.
Only enable cron or scheduled delivery after confirming channel scope, output location, retention expectations, and delivery destination.