The Primer
AdvisoryAudited by Static analysis on Apr 30, 2026.
Overview
No suspicious patterns detected.
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.
Your assistant may push back, slow decisions, or challenge you instead of acting like a neutral helper.
This is disclosed and central to the skill’s coaching purpose, but it intentionally shifts the assistant from simply following requests toward challenging or redirecting the user.
you have permission to be difficult, to notice patterns you'd rather ignore, and to occasionally choose the growth path over the easy path.
Use a lower-friction persona first, and make sure PRIMER.md clearly states when the assistant should back off.
Running the helper can change how the agent behaves in future sessions and may overwrite or alter local workspace files.
The setup helper writes PRIMER.md and mutates AGENTS.md and SOUL.md, which are persistent agent-behavior files, without code-level confirmation, backup, or dry-run controls.
primer_path.write_text(primer_content) ... agents_path.write_text(content) ... soul_path.write_text(content)
Run setup only after reviewing the exact file changes; prefer a dry run, backups, and explicit confirmation before editing AGENTS.md or SOUL.md.
A user may not expect a supposedly instruction-only skill to include a script that edits local agent files.
The package is described as instruction-only, but it includes a runnable setup helper. The helper is visible in the manifest, but the install/runtime expectations are under-declared.
No install spec — this is an instruction-only skill. ... 1 code file(s): scripts/setup_primer.py
Treat setup_primer.py as optional executable code and review it before allowing the agent to run it.
Personal information about your goals, relationships, weaknesses, and self-reflections may remain in workspace files and influence future agent behavior.
The skill stores sensitive personal goals, failure modes, and ongoing reflections in persistent local memory for reuse across sessions.
Patterns to Watch ... Your failure modes. What does it look like when you're at your worst? ... Logged in `memory/YYYY-MM-DD.md` under `## Primer Reflection`
Avoid entering information you would not want stored; periodically review, edit, or delete PRIMER.md and memory logs.
The agent may continue running recurring coaching/reflection workflows after initial setup if those cron jobs are created.
The helper generates recurring scheduled jobs for daily reflection and Miranda check-ins, but the artifacts do not show clear disable, cleanup, or approval boundaries.
"daily_reflection": {"name": "Primer Daily Reflection", "schedule": {"kind": "cron", "expr": "0 7 * * *", "tz": "UTC"}} ... "Cron jobs should be created via the OpenClaw cron tool."Create scheduled jobs only with explicit user approval, document how to pause or remove them, and keep their scope limited to Primer-related files.