ekalavya-self-improvement
v1.0.0Enforce execution discipline for ongoing work after the user has already approved direction. Use when the assistant is at risk of drifting into planning, sta...
⭐ 0· 127·1 current·1 all-time
byJatin Goyal@21f2000735
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
medium confidencePurpose & Capability
The name/description (enforce execution discipline and follow-through) align with the SKILL.md content. The instructions focus on breaking tasks into small execution steps, editing files, verifying results, and committing — all consistent with the stated purpose. There are no unrelated required binaries or environment variables.
Instruction Scope
The instructions explicitly direct the assistant to edit files, run concrete actions, verify results, and commit meaningful changes. They do not ask for unrelated secrets, network exfiltration, or external endpoints. This scope is appropriate for a 'do the work' skill, but it grants the assistant authority to modify user repos/files and to default to 'silent execution mode' (minimal narration). That behavioral authority is intentioned by the skill but is operationally high-impact and should be controlled by environment permissions and user policies.
Install Mechanism
Instruction-only skill with no install spec and no code files. This minimizes disk-write/install risk; there are no downloads or third-party packages to vet.
Credentials
No required environment variables, credentials, or config paths are declared. The SKILL.md references accessing repos and prior work (which implies filesystem/git access), but it does not ask for unrelated credentials or secrets.
Persistence & Privilege
The skill is not force-included (always:false) and doesn't request persistent system privileges. However, it intentionally directs the assistant to autonomously perform edits and commits and to operate silently by default. Because platform agents can run autonomously by default, this behavioral characteristic raises operational risk (unexpected or undesired persistent changes) unless runtime permissions, review workflows, or explicit user approvals are enforced.
Assessment
This skill is coherent for its goal but empowers the assistant to edit files, run actions, and commit changes with minimal narration. Before installing: (1) confirm the agent's runtime/FS permissions (restrict write access to only intended repos/folders); (2) require explicit user approval for destructive or high-impact changes; (3) enable commit-review / CI / backups so changes can be audited and reverted; (4) consider disabling autonomous invocation or adding guardrails if you don't want the assistant to act without catching every change; (5) test in a safe repository/workspace first. If you cannot enforce these operational controls, treat the skill as higher-risk despite being internally coherent.Like a lobster shell, security has layers — review code before you run it.
latestvk97a6wh8gc94c1c6jwee6pjerd8363tx
License
MIT-0
Free to use, modify, and redistribute. No attribution required.