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.
npx agentmods add instructions/ackeskin/contexture/bashgit clone --depth 1 https://github.com/AcKeskin/contextureWhat 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 | $0.00927 | $0.00927 |
| Opus 5 | $0.00464 | $0.00464 |
| Sonnet 5 | $0.00185 | $0.00185 |
| Haiku 4.5 | $0.00093 | $0.00093 |
Grade A, and why
contexture bash.instructions.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 38 lines — stays where its author put it; the contents beside it link to each section on GitHub.
bash rules
Auto-loaded by Copilot when editing files matching
**/*.sh. Generated fromarchitectural-rules/bash/— do not hand-edit.
Bash quoting
Quote every variable expansion: "$var", "${arr[@]}". Unquoted expansions word-split and glob-expand — the source of nearly every "works until a path has a space" bug.
Use "$@" (not $*, not unquoted $@) to forward arguments — it preserves each argument as a separate quoted word. $* joins them into one string.
Use [[ ... ]] over [ ... ] for tests — no word-splitting inside, supports &&/||/=~, and doesn't break on empty variables.
Use $(command) over backticks for command substitution — it nests cleanly and is readable.
⚠ Never use PowerShell here-string syntax (@'...'@) inside a bash command. Bash parses @'text'@ as a literal @, then a single-quoted string, then a trailing @ — so a leading @ leaks silently into the output. A leading @ then silently prefixes the commit subject. For multi-line commit messages in bash use $'line1\nline2', a real here-doc (-F - with <<'EOF'), or -F <file>. Read back the subject (git log -1 --format=%s) after committing to catch it.
Why: unquoted expansion is the single most common shell bug class — it passes every test with simple inputs and breaks the moment a filename has a space or a variable is empty. Source: Google Shell Style Guide, ShellCheck (SC2086). The here-string landmine is sharper on Windows where PowerShell is the other shell in muscle memory — the syntaxes look interchangeable and are not.
Bash safety
Start scripts with set -euo pipefail: -e exit on error, -u error on unset variable, -o pipefail so a failure anywhere in a pipe fails the pipe. Without these, errors are silently ignored and the script charges on.
Set IFS=$'\n\t' when word-splitting matters — the default IFS splits on spaces too and mangles paths/filenames with spaces.
Clean up with trap 'cleanup' EXIT — temp files and locks get removed even on early exit or error. Don't rely on reaching a cleanup line at the bottom.
Verify required commands exist up front (command -v jq >/dev/null || { echo "jq required" >&2; exit 1; }) instead of failing cryptically halfway through.
Terminate only the process you started, by the PID you captured (kill "$pid"). Never pkill / kill / taskkill by name or pattern — name-matching also kills unrelated processes (e.g. pkill godot to stop a headless run also takes down the user's open editor).
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.
- yesterday First seen · 38 lines · 927 tokens per session scan A 16ae548bec91
contexture bash.instructions.md is an instructions file published in the GitHub repository AcKeskin/contexture (2 stars, last pushed 1mo ago), licensed MIT. It adds 927 tokens to every session, about $0.0046 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
agentic-workflow CLAUDE.md
Instructions for gtrabanco/agentic-workflow, covering claude.md, repository layout, working rules, authoring a skill and hand off, don't compose across a model/effort boundary.
ccjk AGENTS.md
Instructions for miounet11/ccjk, covering repository guidelines, source of truth, behavior, validation and notes.
ccjk CLAUDE.md
Instructions for miounet11/ccjk: 持久知识唯一来源:docs/CLAUDE-NOTES.md(请优先阅读本文档).
stenographer-mode copilot-instructions.md
Instructions for AkashAi7/stenographer-mode, covering steno mode — global instructions, contract, rules and levels.
context-engineering AGENTS.md
Instructions for fending/context-engineering, covering agents.md, what this project is, structure, standards and navigation.
capy CLAUDE.md
Claude Code instructions for serpro69/capy, covering claude.md, project, architecture, key packages and critical invariants.