Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/openshift-eng/ai-helpersnpx agentmods add commands/openshift-eng/ai-helpers/pruneWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/commands/openshift-eng/ai-helpers/prune)<a href="https://agentmods.dev/commands/openshift-eng/ai-helpers/prune"><img src="https://agentmods.dev/badge/commands/openshift-eng/ai-helpers/prune/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/commands/openshift-eng/ai-helpers/prune"><img src="https://agentmods.dev/badge/commands/openshift-eng/ai-helpers/prune.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 5.1 | $0.00013 | $0.02085 |
| Opus 5.5 | $0.00005 | $0.00834 |
| Sonnet 5.5 | $0.00003 | $0.00417 |
| Haiku 4.5 | $0.00001 | $0.00209 |
Grade C, and why
prune scanned grade C 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 1mo 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.
Recursive force deletehighDestructive command
rm -rf with a variable or a broad path is one typo away from removing the wrong tree.
git rm -rf plugins/{plugin-name}/ How it starts
The opening of the file, as written. The whole thing — 252 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Name
marketplace-ops:prune
Synopsis
/marketplace-ops:prune [--dry-run]
Description
Analyzes the plugin marketplace for stale content using a two-tier system:
- Plugin-level (mechanical): A scoring script flags entire plugins for removal based on git history and structural signals. If a plugin meets the scoring threshold, it is removed — no LLM override. The script is the decision-maker.
- Item-level (LLM-assisted): For plugins that survive the plugin-level cut, a second script flags individual commands and skills. The LLM then reviews flagged items and applies qualitative judgment (AI reasoning required, duplicates, dead references) to decide which to remove.
By default, creates a branch removing identified items and opens a PR with a removal manifest and /save workflow. With --dry-run, prints the analysis report without taking action.
Arguments
--dry-run: Report pruning candidates without taking any action. No branch, no removals, no PR.
Implementation
Step 1: Run the Plugin Scoring Script
Run the scoring script to get a structured JSON report of all plugins scored by staleness:
python3 plugins/marketplace-ops/scripts/score-plugins.py .
The script handles:
- Reading
.pruneprotectand skipping protected plugins - Inventorying all plugins (commands, skills, hooks, README, version)
- Gathering git metadata (last commit date, commit count, contributor count)
- Detecting batch-update dates (when 5+ plugins share the same last-commit date) and falling through to the second-most-recent commit
- Scoring each plugin against these heuristics (candidate threshold: score >= 3):
| Signal | Weight |
|---|---|
| Last meaningful commit > 90 days ago | 1 |
| Last meaningful commit > 120 days ago | 2 |
| Last meaningful commit > 150 days ago | 3 |
| Last meaningful commit > 180 days ago | 4 |
| Number of commits <= 3 | 2 |
| Number of commits > 3 and <= 5 | 1 |
| Single contributor + inactive > 2 months | 1 |
| Has OWNERS file | -2 |
| Small plugin footprint (few things inside) | 1 |
| Minimal README or docs | 1 |
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.
- 1mo ago First seen · 252 lines · 13 tokens per session scan C d61376d2075a
prune is a command published in the GitHub repository openshift-eng/ai-helpers (121 stars, last pushed today), licensed Apache-2.0. It adds 13 tokens to every session and 2,085 once invoked, about $0.0001 per session on Opus 5.5. A static security scan graded it C with 1 finding (recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-05.
Other commands, from other repositories
gsd:quick
Execute a quick task with GSD guarantees (atomic commits, state tracking) but skip optional agents.
standup
Review my git commits from the past workday and generate a standup summary.
gsd:remove-phase
Remove a future phase from roadmap and renumber subsequent phases.
company-repo
Gestionar el repositorio Company Savia: crear, conectar, estado, sincronizar. El repositorio compartido de la empresa para conocimiento, mensajería y estado.
iterate
Execute a single Ralph loop iteration - complete one work unit from tasks.md with proper commits and progress tracking.
weekly-summary
Generate weekly work summary from git activity.