prompt-safety
Change root or leaf prompts, exposed-method catalogs, history handling, or prompt-budget enforcement without weakening anti-context-rot guarantees.
Skill by williamntlam
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/williamntlam-prompt-safety
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: finance, agent-building
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/prompt-safety
What's inside
- Skill name
prompt-safety- Description
- Change root or leaf prompts, exposed-method catalogs, history handling, or prompt-budget enforcement without weakening anti-context-rot guarantees.
- License
- MIT
- Instructions
- 19 lines in SKILL.md · view on GitHub
Files (2)
- SKILL.md
- prompt-contracts.md · references/
SKILL.md (first lines)
# Prompt safety Use this skill for files in `rlm/prompts/`, `rlm/core/prompt_guard.py`, or `rlm/core/history.py`, and for behavior that changes what reaches a model. ## Why it matters The product's value depends on a parent model that reasons from compact, observable findings rather than a growing transcript or copied source dump. The token and instruction limits protect reliability as well as cost. Read [`references/prompt-contracts.md`](references/prompt-contracts.md) before changing limits, routing, or history. Inspect the narrow prompt or guard test before editing; load `../testing/SKILL.md` to run it. Keep generic prompts concise. New exposed REPL or domain methods affect the instruction count and must be represented in the prompt catalog.
From williamntlam/recursive-language-model, MIT.
More from williamntlam/recursive-language-model
- 0.526score
evaluation
williamntlam
Add or modify RLM evaluations, including Harbor agent tasks, LLM-as-judge utilities, rubrics, cases, or long-context benchmarks while keeping model cost…
SkillBFreshness: BClaude Code +9 - 0.526score
cli-and-config
williamntlam
Change CLI commands, flags, TOML/YAML configuration, precedence, or startup errors while keeping authentication out of config files.
SkillBFreshness: BClaude Code +9 - 0.526score
docker-repl
williamntlam
Change the Docker REPL image, in-container namespace, RPC boundary, or isolation settings without exposing source writes, credentials, or network access.
SkillBFreshness: BClaude Code +9 - 0.526score
runtime-development
williamntlam
Change RLM runtime behavior, repository or corpus domains, backends, or the Python API while preserving data isolation and budget behavior.
SkillBFreshness: BClaude Code +9 - 0.521score
observability
williamntlam
Change trajectory logging, report generation, or runtime error reporting while preserving useful diagnostics and safe HTML output.
SkillBFreshness: BClaude Code +9 - 0.521score
overengineering-check
williamntlam
Review an RLM specification, design, or change for complexity without a clear justification, including speculative abstraction and premature extensibility.
SkillBFreshness: BClaude Code +9 - 0.496score
testing
williamntlam
Add, update, or diagnose pytest coverage for RLM behavior using focused fake environments and optional Docker integration tests.
SkillBFreshness: BClaude Code +9
Signals
- Stars
- 0
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Aug 23, 2026 (2 months ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from williamntlam
- 0.526score
evaluation
williamntlam
Add or modify RLM evaluations, including Harbor agent tasks, LLM-as-judge utilities, rubrics, cases, or long-context benchmarks while keeping model cost…
SkillBFreshness: BClaude Code +9 - 0.526score
cli-and-config
williamntlam
Change CLI commands, flags, TOML/YAML configuration, precedence, or startup errors while keeping authentication out of config files.
SkillBFreshness: BClaude Code +9 - 0.526score
docker-repl
williamntlam
Change the Docker REPL image, in-container namespace, RPC boundary, or isolation settings without exposing source writes, credentials, or network access.
SkillBFreshness: BClaude Code +9 - 0.526score
runtime-development
williamntlam
Change RLM runtime behavior, repository or corpus domains, backends, or the Python API while preserving data isolation and budget behavior.
SkillBFreshness: BClaude Code +9 - 0.521score
observability
williamntlam
Change trajectory logging, report generation, or runtime error reporting while preserving useful diagnostics and safe HTML output.
SkillBFreshness: BClaude Code +9 - 0.521score
overengineering-check
williamntlam
Review an RLM specification, design, or change for complexity without a clear justification, including speculative abstraction and premature extensibility.
SkillBFreshness: BClaude Code +9
Cite this page
prompt-safety skill. Agent Mart. https://agentmart.dev/skills/williamntlam-prompt-safety (last updated 2026-10-24)