What is finesse-execute?

A command for starting a Finesse-managed ralph loop from a saved plan or a prompt. The loop repeatedly runs an agent workflow and tracks its progress.

Use it to launch a prepared Finesse plan or start a loop from an inline prompt.

It handles plan discovery, setup, snapshots, and per-iteration records so a long-running workflow can be executed and reviewed.

finesse-execute — Command by jonathanung

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods

npx agentmods add commands/jonathanung/finesse/finesse-execute

Clone the repo

git clone --depth 1 https://github.com/jonathanung/finesse

Per session 15 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

When invoked 1,514 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 1 finding. Scan, not verified.

Origin original No closer match found in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5	$0.00015	$0.01514
Opus 5	$0.00008	$0.00757
Sonnet 5	$0.00003	$0.00303
Haiku 4.5	$0.00002	$0.00151

Measured 2d ago against content hash d12370caf2d2, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

finesse-execute scanned grade A with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Asks the agent to reveal its instructionslowSystem prompt leakage

Directions to print, repeat or translate the system prompt extract configuration the operator did not intend to expose.

- Ask: "Execute this plan? [Yes / No / Show full prompt]"

Downgraded: this mod is about security review, or the phrase is quoted, so it is likely naming the pattern rather than instructing it.

plugins/finesse/commands/finesse-execute.md · 129 lines

How it starts

The opening of the file, as written. The whole thing — 129 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Finesse Execute

Launch a Finesse-managed ralph loop. This is a forked execution layer based on the ralph-wiggum plugin, enhanced with file-based input, pre-execution git snapshots, and per-iteration telemetry for retro analysis.

Argument Parsing

Parse $ARGUMENTS for these patterns:

Pattern 1: Direct plan reference (most common after /finesse)

If $ARGUMENTS contains --prompt-file, pass all arguments directly to the setup script:

python3 "${CLAUDE_PLUGIN_ROOT}/scripts/finesse_execute.py" $ARGUMENTS

Pattern 2: Auto-detect from finesse-plans/

If $ARGUMENTS is empty or blank:

Use Glob to scan finesse-plans/ for plan files:
- Single-workflow: finesse-plans/*-promise.txt (paired with finesse-plans/*.md)
- Multi-workflow: finesse-plans/*/execution-graph.md
If no plans found, say "No plans found in finesse-plans/. Run /finesse first to generate a plan, or provide a prompt inline." and stop.
If exactly one single-workflow plan found:
- Read the -plan.md file to extract task type, approach summary, and iteration count
- Present a summary:
```
Found plan: <plan-name>
Task type: <type>
Max iterations: <N>
Prompt: <first 3 lines of prompt file>...
Estimated cost: <range from plan metadata>
```
- Ask: "Execute this plan? [Yes / No / Show full prompt]"
- On Yes: execute with --prompt-file and --completion-promise-file pointing to the plan files
- On Show full prompt: display the full prompt, then ask again
If exactly one multi-workflow session found:
- Read execution-graph.md to show the wave structure
- Present wave-by-wave summary
- Ask: "Execute this multi-workflow plan? [Yes / No]"
- On Yes: execute the waves orchestrator directly via Bash:
```
python3 "${CLAUDE_PLUGIN_ROOT}/scripts/finesse_waves.py" start <session-name>
```
- Stop.
If multiple plans found:
- List each plan with its name, task type, and iteration count (read from -plan.md or plan.md)
- Ask the user to pick one using AskUserQuestion
- Proceed with the selected plan as in step 3

Read the full file on GitHub · 129 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

2d ago First seen · 129 lines · 15 tokens per session scan A d12370caf2d2

Subscribe to this mod's changes

finesse-execute is a command published in the GitHub repository jonathanung/finesse (4 stars, last pushed 6mo ago), licensed MIT. It adds 15 tokens to every session and 1,514 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 1 finding (asks the agent to reveal its instructions). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Building agents

finesse-execute

Getting it into your agent

What it costs to keep this loaded

Grade A, and why

Asks the agent to reveal its instructionslowSystem prompt leakage

How it starts

Finesse Execute

Argument Parsing

Pattern 1: Direct plan reference (most common after /finesse)

Pattern 2: Auto-detect from finesse-plans/

What this file has done since we first saw it

Other commands, from other repositories

claude-flow-swarm

bootstrap

ox-session-review

add-agent

handoff

xdo