CB install-memex-claw
Wire memex (the local-first MCP memory server) into an OpenClaw gateway — works wherever OpenClaw runs (Linux or macOS, VPS or workstation, self-hosted OpenClaw or Moonshot Kimi-Claw). Installs the memex-mvp daemon that auto-captures every OpenClaw session in ~/.openclaw/agents/main/sessions/ into a single SQLite + FTS5 corpus, and merges memex into the gateway's MCP-server config so the LLM gets 11 search/retrieval tools (memex_search, memex_recent, memex_overview, etc.). Auto-detects whether memex-mvp is already installed (e.g. via the generic install-memex skill for Claude Code) — if yes, skips install and just wires OpenClaw + back-fills past sessions; if no, does the full platform-aware install (Linux → systemd user-service, macOS → LaunchAgent). Zero questions to the user — discovery → actions → verification. Use when the user says "set up memex for OpenClaw", "wire memex into my OpenClaw", "make OpenClaw remember its sessions", "поставь memex здесь", or similar. PAIRS with the generic install-memex skill — if the user ALSO uses Claude Code, Cursor, Cline, Continue, or Zed on the same machine, recommend they run install-memex separately for those flows.
As a process B 71/100 · Nearly there — weak spots: result and completion
What is at stake
Medium-severity findings: the skill is probably honest, but read what alarmed the scanner.
Below is the worst case for this category. The finding here is medium: the guard saw a sign, not a proof.
The skill contains commands that delete files, rewrite disks or execute code fetched from the network. The agent may run them without asking if it believes the instructions require it.
Replace destructive commands with safe equivalents that ask for confirmation, scope them to one folder, and stop piping curl into a shell: pin a version and a checksum.
How to improve
- Shorten the description to 1024 characters.
- Your own cases (evals/evals.json, 4–6 real requests with expected answers): the full check would then run those instead of a model-drafted suite.
- A spec.yaml with trigger phrases and assertions — a behaviour contract for CI; `skilltest init` writes a template.
Guard findings · 4
✓ No critical or high findings
Medium and low: 4
-
medium Dangerous commands
cmd-persistenceSKILL.md:44Persistence mechanism (cron / launchd / scheduled task / autorun registry) (documentation table row)| `memex-sync install` | Register the auto-capture daemon — systemd user-service on Linux, LaunchAgent on macOS. The daemon watches `~/.openclaw/agents/main/sessions/` and writes to `~/.memex/data/mem
table -
medium Dangerous commands
cmd-shell-rcSKILL.md:162Writes to a shell startup fileecho 'export PATH=$HOME/.npm-global/bin:$PATH' >> ~/.zshrc 2>/dev/null
-
medium Dangerous commands
cmd-shell-rcSKILL.md:163Writes to a shell startup fileecho 'export PATH=$HOME/.npm-global/bin:$PATH' >> ~/.bashrc 2>/dev/null
-
medium Dangerous commands
cmd-persistenceSKILL.md:186Persistence mechanism (cron / launchd / scheduled task / autorun registry) (documentation table row)| **macOS** (`uname -s` = Darwin) | Writes `~/Library/LaunchAgents/com.parallelclaw.memex.sync.plist`, runs `launchctl load`. Daemon auto-starts on every login. |
table
Files scanned: 2. Evidence is masked. Grey chips explain why severity was lowered.
Against the Agent Skills spec
- error
description-longdescription is 1177 chars, limit 1024
Process rating: all ten parameters 71/100
- 0Result and completion. Does not say what the result is
- 60Tools and files. Uses tools (bash, node) that frontmatter does not declare
- 70Inputs and preconditions. Inputs and preconditions are listed
- 70Execution cost. Instruction body is 4937 tokens
- 85Steps. 29 steps, 1 vague phrases
- 100When it triggers. States when to use and when not to
- 100Failures and branches. 8 branches, has a failure section
- 100Consistency. Name and required fields are in place
- 100Running it twice. Mutating operations check current state
- 100Progress reporting. Reports progress
- medium Safety rules and hard prohibitions inside a skill: they belong in the system prompt, here they protect nothing
- low 16 top-level sections: this looks like several domains in one skill
- low The response is described with custom markup (3 tags): a typed call is more reliable
Everything here is measured from the skill text rather than judged by a model, so the numbers are checkable. A parameter weighs more when it is a more common reason for the process to stall.
Quality signals
- +4Description does not say when NOT to use the skill (false activations)
- +3Description length 1177: 120–800 characters recommended
- +3Output format is not stated: the model decides each time
- -4Absolute local paths (C:\Users, /home/…): not portable
- +1No license
- +2Single-language instructions
- +5Description quotes 4 example trigger phrases
- +4Structure: 23 headings
- +3Step-by-step instructions: 29 items
- +4Has examples (19 code blocks)
Quality base 70; lint remarks subtract, signals add up to 100. Result: 57.