ClawForage Prompt Optimizer

Analyzes your conversation transcripts daily to find patterns, suggest SOUL.md improvements, and recommend skills

Security Scan

The skill's files, runtime instructions, and required binaries are coherent with its stated purpose (analyzing transcripts and proposing SOUL.md improvements); no unexpected network activity or unrelated credentials are requested — but review and approve read-access to your transcripts and note minor operational mismatches documented below.

Purpose & Capability

The skill claims to analyze conversation transcripts and suggest SOUL.md changes and skills. The included scripts and templates directly implement that: extract-transcripts.sh parses JSONL session files, templates/weekly-report.md defines the report shape, and validate-report.sh checks for required sections. Required binaries (jq, bash) are appropriate and expected.

Instruction Scope

The runtime instructions explicitly read user transcripts (~/.openclaw/agents/default/sessions/) and memory/SOUL.md and write reports to memory/optimization/day-{DATE}.md — this is consistent with the stated purpose but does involve reading sensitive local data (transcripts and agent configuration). Two operational inconsistencies to note: (1) SKILL.md says to create a daily report but validate-report.sh looks for the literal header '# ClawForage Weekly Optimization' (a weekly header) — reports that do not include that exact header will fail validation; (2) SKILL.md uses a {baseDir} placeholder when invoking scripts (bash {baseDir}/scripts/...), so the agent runtime must substitute baseDir correctly or the script call may fail. Also, extract-transcripts.sh concatenates all files into one shell variable (ALL_MESSAGES) which can be memory-heavy on very large archives — operational risk, not malicious.

Install Mechanism

There is no install spec that downloads or executes remote code; the skill is delivered with local scripts and templates. That reduces supply-chain risk. The code files are plain shell scripts and a template; nothing is fetched from external URLs during install.

Credentials

The skill requests no environment variables or credentials. It does require access to local transcript files and memory/SOUL.md which is proportionate to its purpose. No unrelated secrets or config paths are requested.

Persistence & Privilege

always:false (not force-included) and disable-model-invocation:false (normal autonomous invocation allowed). The skill does not request system-wide config changes or the modification of other skills. It writes its own report files under memory/optimization as intended.

Assessment

This skill appears internally consistent with its goal of analyzing transcripts and suggesting SOUL.md improvements. Before installing, consider: 1) Do you want a skill reading your local transcripts and SOUL.md? These contain sensitive content; grant only to skills you trust. 2) Ensure jq and bash are installed and that the agent runtime will substitute {baseDir} correctly (or edit the invocation to an absolute path). 3) Verify the report format includes the literal header '# ClawForage Weekly Optimization' (or update validate-report.sh) so validation succeeds. 4) If you have large transcript archives, the extractor may use lots of memory — test on a sample. If these points are acceptable, the skill is coherent and suitable to install.

Prompt & Workflow Optimizer

You are a meta-analysis agent run by ClawForage. Your job: review the user's recent conversation transcripts and produce an actionable daily optimization report.

Step 1: Extract Transcript Data

Run the extraction script on the user's transcripts directory:

bash {baseDir}/scripts/extract-transcripts.sh ~/.openclaw/agents/default/sessions/ 1

This outputs a structured summary of:

• All user questions from the past day
• Repeated questions (exact matches)
• Tool usage frequency
• Failures and errors
• Cost summary

Read the output carefully before proceeding.

Step 2: Read Current SOUL.md

cat memory/SOUL.md 2>/dev/null || echo "No SOUL.md found"

Understand the user's current agent configuration so you can suggest meaningful improvements.

Step 3: Analyze and Write Report

Based on the extracted data and current SOUL.md, write a report to memory/optimization/day-{DATE}.md where {DATE} is today's date in YYYY-MM-DD format.

Create the directory first:

mkdir -p memory/optimization

Your report MUST follow this structure (use {baseDir}/templates/weekly-report.md as reference):

Repeated Patterns

Identify questions asked 2+ times. For each:

• State the pattern and frequency
• Suggest a concrete action: add info to SOUL.md, create a cron job, or install a skill

SOUL.md Suggestions

Propose specific additions or changes to SOUL.md. Write them as ready-to-copy text blocks. Examples:

• Adding default preferences ("Default weather location: Hangzhou")
• Adding workflow shortcuts ("When asked to translate, always use DeepL API first")
• Removing outdated instructions

Recommended Skills

Based on the user's most common tasks, suggest relevant skills. For each:

• Skill name and what it does
• Why it matches the user's usage pattern
• Install command: openclaw skill install <name>

Failure Analysis

For each error or multi-attempt task:

• What went wrong
• Root cause (missing dependency, unclear prompt, wrong tool)
• Suggested prevention (add to SOUL.md, install dependency, create pre-check skill)

Usage Stats

Summarize: message count, total cost, average cost, top tools, topic distribution.

Step 4: Validate Report

bash {baseDir}/scripts/validate-report.sh memory/optimization/day-{DATE}.md

If validation fails, fix the missing sections and re-validate.

Constraints

• Read-only: Never modify transcripts, SOUL.md, or any existing files
• Suggestions only: Present changes for user approval, never auto-apply
• Concise: Max 500 words per report
• Model: Uses your default configured model — no override needed
• Privacy: Never include full message content in reports — summarize patterns only