oracle

Best practices for using the oracle CLI (prompt + file bundling, engines, sessions, and file attachment patterns).

1.70x

Quality

70%

Does it follow best practices?

Impact

99%

1.70x

Average score across 3 eval scenarios

Securityby

Advisory

Suggest reviewing before use

Optimize this skill with Tessl

npx tessl skill review --optimize ./openclaw/skills/oracle/SKILL.md

Evaluation results

100%

48%

Code Review Automation Script

File selection and preview workflow

Criteria

Without context

With context

Dry-run preview step

100%

Files-report flag

100%

Browser engine

100%

GPT-5.2 Pro model

100%

Slug for session readability

100%

Tight file set (src only)

100%

Negative glob excludes tests

50%

100%

Negative glob excludes snapshots or build artifacts

50%

100%

No secrets attached

100%

Script is executable/runnable

87%

100%

Avoids pnpx for install fallback

100%

Without context: $0.3219 · 1m 21s · 21 turns · 22 in / 4,598 out tokens

With context: $0.2458 · 1m 3s · 15 turns · 261 in / 2,805 out tokens

100%

62%

Oracle Session Runbook for Long-Running Investigations

Session management and reattachment

Criteria

Without context

With context

No re-run on timeout

66%

100%

Status listing command

30%

100%

Session attach command

33%

100%

Slug on initial run

100%

Force flag purpose explained

100%

Session storage location

100%

ORACLE_HOME_DIR override

100%

One-shot run nature noted

10%

100%

Restore context method

40%

100%

Workflow document produced

100%

Without context: $0.4747 · 2m 14s · 22 turns · 3,371 in / 7,514 out tokens

With context: $0.3966 · 1m 39s · 20 turns · 425 in / 5,099 out tokens

97%

11%

Production Debugging Investigation with Oracle

Prompt crafting and safety

Criteria

Without context

With context

Project briefing included

60%

70%

Where things live

80%

100%

Exact question and error text

100%

What was already tried

87%

100%

Constraints stated

62%

100%

Desired output format

87%

100%

No secrets in file list

100%

Briefing length for exhaustive prompt

75%

100%

All context files listed

87%

100%

Assumes zero model memory

100%

Reproducible prompt artifact

100%

Without context: $0.2298 · 1m 7s · 13 turns · 16 in / 3,505 out tokens

With context: $0.3320 · 1m 16s · 17 turns · 21 in / 4,050 out tokens

Repository: trpc-group/trpc-agent-go
Commit: 8763418

Evaluated: 16 days ago
Agent: Claude Code
Model: Claude Sonnet 4.6

Table of Contents

Code Review Automation Script Oracle Session Runbook for Long-Running Investigations Production Debugging Investigation with Oracle

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.