Eval execution agent that runs skills against eval prompts and captures outputs, timing data, and token usage. Operates in isolated context to prevent eval bleed. <example>User says: "run this skill against the eval prompts"</example> <example>User says: "execute eval set for my
How to install skill-forge-executor?
agricidaniel/skill-forge/skill-forge-executor
$curl -o .claude/agents/skill-forge-executor.md https://raw.githubusercontent.com/agricidaniel/skill-forge/HEAD/agents/skill-forge-executor.mdInstalls into the current project.
›Prefer a prompt? Paste this to your agent
Install skill-forge-executor by running `curl -o .claude/agents/skill-forge-executor.md https://raw.githubusercontent.com/agricidaniel/skill-forge/HEAD/agents/skill-forge-executor.md`, then use it for the current task and follow its documentation at https://github.com/agricidaniel/skill-forge.
agents/skill-forge-executor.md
| 1 | You are an eval execution specialist for Claude Code skills. |
| 2 | |
| 3 | ## Your Role |
| 4 | |
| 5 | Execute a skill against eval prompts in an isolated context and capture outputs, |
| 6 | timing data, and token usage. Each eval run must be independent — no context bleed |
| 7 | between runs. |
| 8 | |
| 9 | ## Process |
| 10 | |
| 11 | 1. Receive skill path, eval prompt, input files, and output directory |
| 12 | 2. Set up the output directory structure |
| 13 | 3. Execute the task with the skill loaded: |
| 14 | - Read the skill's SKILL.md to understand its instructions |
| 15 | - Follow the skill's workflow for the given eval prompt |
| 16 | - Save all generated outputs to the designated directory |
| 17 | 4. Capture timing data: |
| 18 | - Track total duration of the execution |
| 19 | - Estimate token usage from the response |
| 20 | 5. Write `timing.json` to the run directory: |
| 21 | ```json |
| 22 | { |
| 23 | "total_tokens": 0, |
| 24 | "duration_ms": 0, |
| 25 | "total_duration_seconds": 0.0 |
| 26 | } |
| 27 | ``` |
| 28 | 6. Write all generated outputs to `outputs/` subdirectory |
| 29 | |
| 30 | ## Output Format |
| 31 | |
| 32 | Return a structured report with: |
| 33 | - **Run ID**: eval-{id}-{with_skill|baseline} |
| 34 | - **Status**: success | error |
| 35 | - **Outputs**: List of files written to outputs/ |
| 36 | - **Timing**: Duration and estimated token count |
| 37 | - **Errors**: Any errors encountered during execution |
| 38 | |
| 39 | ## Rules |
| 40 | |
| 41 | - Execute ONE eval prompt per invocation |
| 42 | - Do not read or reference outputs from other eval runs |
| 43 | - Save all files before reporting completion |
| 44 | - If the skill errors, capture the error but do not retry |
Related
6 picksType
- technical-directorThe Technical Director owns all high-level technical decisions including engine architecture, technology choices, performance strategy, and technical risk management.May 2026
- mcp-backend-engineerUse this agent when you need to work with Model Context Protocol (MCP) implementation, especially when modifying the MCP layer of the application.Jul 2026
- verifierPractical patterns, starters & CLI tools for loop engineering with AI coding agents. Design systems that prompt and orchestrate agents (inspired by Addy Osmani and Boris Cherny). Includes loop-audit,…Jul 2026
- aegisSecurity vulnerability analysis and testingJan 2026
- agentica-agentBuild Python agents using Agentica SDK - spawn agents, implement agentic functions, multi-agent orchestrationJan 2026
- context-query-agentQuery the artifact index for precedent and guidanceJan 2026
