Borrowing it
Nothing to install: this file belongs to justomsharma/github-resume-assistant. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/justomsharma/github-resume-assistant/main/.claude/skills/prompt-practices/SKILL.mdgit clone --depth 1 https://github.com/justomsharma/github-resume-assistantWrote 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/skills/justomsharma/github-resume-assistant/prompt-practices)<a href="https://agentmods.dev/skills/justomsharma/github-resume-assistant/prompt-practices"><img src="https://agentmods.dev/badge/skills/justomsharma/github-resume-assistant/prompt-practices/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/skills/justomsharma/github-resume-assistant/prompt-practices"><img src="https://agentmods.dev/badge/skills/justomsharma/github-resume-assistant/prompt-practices.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 | $0.00066 | $0.01023 |
| Opus 5.5 | $0.00026 | $0.00409 |
| Sonnet 5.5 | $0.00013 | $0.00205 |
| Haiku 4.5 | $0.00007 | $0.00102 |
Grade A, and why
prompt-practices 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 2mo 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.
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 — 81 lines — stays where its author put it; the contents beside it link to each section on GitHub.
prompt-practices
The rule: every prompt we send to Claude is code. Write it to be read, grounded, and testable — never a vibe.
When you write or change any prompt string that goes to the Anthropic API in this
project, follow this. It pairs with docs/CODING_PRACTICES.md (the API-clients and
model-id rules) and the claude-api skill (SDK usage, model ids, params).
Where prompts live (non-negotiable)
- Prompts are assembled in
clients/anthropic.py, one function per tool's prompt (ARCHITECTURE.md: clients own prompt assembly).core/never builds prompts. - The model id comes from
config.anthropic_model, never hardcoded in the prompt or the call. - Keep the prompt text in a named module-level constant or a small builder function so a test can assert what we send.
The seven rules for a good prompt here
-
Assign a role + task in one line. Start the system prompt with who Claude is and the single job. "You extract concrete, verifiable claims from an engineer's resume." One responsibility per prompt — mirror the "functions do one thing" rule.
-
Ground every instruction in the real input. Our whole moat is grounding in real data (PRODUCT.md). Tell Claude to use only the supplied resume/GitHub text and to never invent facts not present in the input.
-
Separate instructions from data. Put the user's resume text inside clear delimiters (XML tags like
<resume>...</resume>), so injected text in the resume can't be read as instructions. Never string-concatenate user text into the middle of an instruction sentence. -
Demand structured output and specify the schema. Ask for JSON matching an exact shape (field names, types), and say "return only the JSON, no prose." This is what
core/parses intoClaimdataclasses — the shape is a contract, so state it explicitly and keep it in sync withcore/models.py. -
Handle the empty / thin case in the prompt itself. Our real user has a near-empty resume or GitHub (PRODUCT.md). Tell Claude what to do when there's little to extract: return an empty list, not fabricated claims. Never let the prompt pressure Claude into inventing content to "find something."
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.
- 2mo ago First seen · 81 lines · 66 tokens per session scan A eaa35317480d
prompt-practices is a skill published in the GitHub repository justomsharma/github-resume-assistant (0 stars, last pushed 3mo ago), licensed MIT. It adds 66 tokens to every session and 1,023 once invoked, about $0.0003 per session on Opus 5.5. 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 skills, from other repositories
gigachat-chat
A chat command that sends a user prompt to GigaChat, an AI text-generation service, through MCP, a way for tools to connect to an agent. It can pass a model and temperature and reports the model and token counts.
dspy
DSPy: declarative LM programs, auto-optimize prompts, RAG.
claude-api
Reference for the Claude API / Anthropic SDK — model ids, pricing, params, streaming, tool use, MCP, agents, caching, token counting, model migration. TRIGGER — read BEFORE opening the target file; don't skip because it "looks like a one-liner" — whenever: the prompt names Claude/Anthropic in any form (Claude…
omni-compression
Configure RTK (command output), Caveman (prose), and stacked compression modes. Manage language packs, custom rules, and test prompt compression reducing tokens by 60–90%.
cli-compression
Configure and test prompt compression from the CLI. Manage RTK filters, Caveman rules, stacked compression modes, and preview compression output with real prompts.
claude-api-in-prototypes
Call Claude from your HTML artifacts via window.claude.complete.